d/freebsd-nq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
f4c01f1508
freebsd-nq/sys/kern/sched_ule.c

1989 lines
50 KiB
C
Raw Normal View History

- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
/*-
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
* Copyright (c) 2002-2003, Jeffrey Roberson <jeff@freebsd.org>
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
* 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 unmodified, 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 ``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 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.
*/
Use __FBSDID().
2003-06-11 00:56:59 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
Turn PREEMPTION into a kernel option. Make sure that it's defined if FULL_PREEMPTION is defined. Add a runtime warning to ULE if PREEMPTION is enabled (code inspired by the PREEMPTION warning in kern_switch.c). This is a possible MT5 candidate.
2004-09-02 18:59:15 +00:00
#include <opt_sched.h>
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
#define kse td_sched
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
#include <sys/param.h>
#include <sys/systm.h>
Update for the KDB framework: o Call kdb_backtrace() instead of backtrace().
2004-07-10 21:38:22 +00:00
#include <sys/kdb.h>
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
#include <sys/kernel.h>
#include <sys/ktr.h>
#include <sys/lock.h>
#include <sys/mutex.h>
#include <sys/proc.h>
- Add in support for KSEs with 0 slice values on the run queue. If we try to select a KSE with a slice of 0 we will update its slice and insert it onto the next queue. - Pass the KSE instead of the ksegrp into sched_slice(). This more accurately reflects the behavior of the code. Slices are granted to kses. - Add a function kseq_nice_min() which finds the smallest nice value assigned to the kseg of any KSE on the queue. - Rewrite the logic in sched_slice(). Add a large comment describing the new slice selection scheme. To summarize, slices are assigned based on the nice value. Priorities are still calculated based on the nice and interactivity of a process. Slice sizes of 0 may be granted for KSEs whos nice is 20 or futher away from the lowest nice on the run queue. Other nice values are scaled across the range [min, min+20]. This fixes ULEs bad behavior with positively niced processes.
2003-04-02 06:46:43 +00:00
#include <sys/resource.h>
- Add initial support for pinning and binding.
2003-11-04 07:45:41 +00:00
#include <sys/resourcevar.h>
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
#include <sys/sched.h>
#include <sys/smp.h>
#include <sys/sx.h>
#include <sys/sysctl.h>
#include <sys/sysproto.h>
Rework the interface between priority propagation (lending) and the schedulers a bit to ensure more correct handling of priorities and fewer priority inversions: - Add two functions to the sched(9) API to handle priority lending: sched_lend_prio() and sched_unlend_prio(). The turnstile code uses these functions to ask the scheduler to lend a thread a set priority and to tell the scheduler when it thinks it is ok for a thread to stop borrowing priority. The unlend case is slightly complex in that the turnstile code tells the scheduler what the minimum priority of the thread needs to be to satisfy the requirements of any other threads blocked on locks owned by the thread in question. The scheduler then decides where the thread can go back to normal mode (if it's normal priority is high enough to satisfy the pending lock requests) or it it should continue to use the priority specified to the sched_unlend_prio() call. This involves adding a new per-thread flag TDF_BORROWING that replaces the ULE-only kse flag for priority elevation. - Schedulers now refuse to lower the priority of a thread that is currently borrowing another therad's priority. - If a scheduler changes the priority of a thread that is currently sitting on a turnstile, it will call a new function turnstile_adjust() to inform the turnstile code of the change. This function resorts the thread on the priority list of the turnstile if needed, and if the thread ends up at the head of the list (due to having the highest priority) and its priority was raised, then it will propagate that new priority to the owner of the lock it is blocked on. Some additional fixes specific to the 4BSD scheduler include: - Common code for updating the priority of a thread when the user priority of its associated kse group has been consolidated in a new static function resetpriority_thread(). One change to this function is that it will now only adjust the priority of a thread if it already has a time sharing priority, thus preserving any boosts from a tsleep() until the thread returns to userland. Also, resetpriority() no longer calls maybe_resched() on each thread in the group. Instead, the code calling resetpriority() is responsible for calling resetpriority_thread() on any threads that need to be updated. - schedcpu() now uses resetpriority_thread() instead of just calling sched_prio() directly after it updates a kse group's user priority. - sched_clock() now uses resetpriority_thread() rather than writing directly to td_priority. - sched_nice() now updates all the priorities of the threads after the group priority has been adjusted. Discussed with: bde Reviewed by: ups, jeffr Tested on: 4bsd, ule Tested on: i386, alpha, sparc64
2004-12-30 20:52:44 +00:00
#include <sys/turnstile.h>
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
#include <sys/vmmeter.h>
#ifdef KTRACE
#include <sys/uio.h>
#include <sys/ktrace.h>
#endif
#include <machine/cpu.h>
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
#include <machine/smp.h>
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
/* decay 95% of `p_pctcpu' in 60 seconds; see CCPU_SHIFT before changing */
/* XXX This is bogus compatability crap for ps */
static fixpt_t ccpu = 0.95122942450071400909 * FSCALE; /* exp(-1/20) */
SYSCTL_INT(_kern, OID_AUTO, ccpu, CTLFLAG_RD, &ccpu, 0, "");
static void sched_setup(void *dummy);
SYSINIT(sched_setup, SI_SUB_RUN_QUEUE, SI_ORDER_FIRST, sched_setup, NULL)
Clean up whitespace, increase consistency and correctness. Submitted by: bde
2004-07-23 23:09:00 +00:00
static SYSCTL_NODE(_kern, OID_AUTO, sched, CTLFLAG_RW, 0, "Scheduler");
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
Clean up whitespace, increase consistency and correctness. Submitted by: bde
2004-07-23 23:09:00 +00:00
SYSCTL_STRING(_kern_sched, OID_AUTO, name, CTLFLAG_RD, "ule", 0,
"Scheduler name");
Add the sysctl node 'kern.sched.name' that has the name of the scheduler currently in use. Move the 4bsd kern.quantum node to kern.sched.quantum for consistency.
2004-06-21 22:05:46 +00:00
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
static int slice_min = 1;
SYSCTL_INT(_kern_sched, OID_AUTO, slice_min, CTLFLAG_RW, &slice_min, 0, "");
- Fix the maximum slice value. I accidentally checked in a value of '2' which meant no process would run for longer than 20ms. - Slightly redo the interactivity scorer. It follows the same algorithm but in a slightly more correct way. Previously values above half were incorrect. - Lower the interactivity threshold to 20. It seems that in testing non- interactive tasks are hardly ever near there and expensive interactive tasks can sometimes surpass it. This area needs more testing. - Remove an unnecessary KTR. - Fix a case where an idle thread that had an elevated priority due to priority prop. would be placed back on the idle queue. - Delay setting NEEDRESCHED until userret() for threads that haad their priority elevated while in kernel. This gives us the same context switch optimization as SCHED_4BSD. - Limit the child's slice to 1 in sched_fork_kse() so we detect its behavior more quickly. - Inhert some of the run/slp time from the child in sched_exit_ksegrp(). - Redo some of the priority comparisons so they are more clear. - Throttle the frequency of sched_pctcpu_update() so that rounding errors do not make it invalid.
2003-06-15 02:18:29 +00:00
static int slice_max = 10;
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
SYSCTL_INT(_kern_sched, OID_AUTO, slice_max, CTLFLAG_RW, &slice_max, 0, "");
int realstathz;
int tickincr = 1;
- Enable STRICT_RESCHED until code that dynamically decides on resched strictness based on the current workload is finished.
2003-02-10 14:11:23 +00:00
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
/*
* The schedulable entity that can be given a context to run.
* A process may have several of these. Probably one per processor
* but posibly a few more. In this universe they are grouped
* with a KSEG that contains the priority and niceness
* for the group.
*/
struct kse {
TAILQ_ENTRY(kse) ke_procq; /* (j/z) Run queue. */
int ke_flags; /* (j) KEF_* flags. */
struct thread *ke_thread; /* (*) Active associated thread. */
fixpt_t ke_pctcpu; /* (j) %cpu during p_swtime. */
char ke_rqindex; /* (j) Run queue index. */
enum {
KES_THREAD = 0x0, /* slaved to thread state */
KES_ONRUNQ
} ke_state; /* (j) thread sched specific status. */
int ke_slptime;
int ke_slice;
struct runq *ke_runq;
u_char ke_cpu; /* CPU that we have affinity for. */
/* The following variables are only used for pctcpu calculation */
int ke_ltick; /* Last tick that we were running on */
int ke_ftick; /* First tick that we were running on */
int ke_ticks; /* Tick count */
};
#define td_kse td_sched
#define td_slptime td_kse->ke_slptime
#define ke_proc ke_thread->td_proc
#define ke_ksegrp ke_thread->td_ksegrp
/* flags kept in ke_flags */
#define KEF_SCHED0 0x00001 /* For scheduler-specific use. */
#define KEF_SCHED1 0x00002 /* For scheduler-specific use. */
#define KEF_SCHED2 0x00004 /* For scheduler-specific use. */
#define KEF_SCHED3 0x00008 /* For scheduler-specific use. */
- In kseq_choose(), don't recalculate slice values for processes with a nice of 0. Doing so can cause an infinite loop because they should be running, but a nice -20 process could prevent them from doing so. - Add a new flag KEF_PRIOELEV to flag a thread that has had its priority elevated due to priority propagation. If a thread has had its priority elevated, we assume that it must go on the current queue and it must get a slice. - In sched_userret() if our priority was elevated and we shouldn't have a timeslice, yield here until we should. Found/Tested by: glebius
2004-12-14 10:34:27 +00:00
#define KEF_SCHED4 0x00010
#define KEF_SCHED5 0x00020
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
#define KEF_DIDRUN 0x02000 /* Thread actually ran. */
#define KEF_EXIT 0x04000 /* Thread is being killed. */
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
/*
* These datastructures are allocated within their parent datastructure but
* are scheduler specific.
*/
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
#define ke_assign ke_procq.tqe_next
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
#define KEF_ASSIGNED 0x0001 /* Thread is being migrated. */
#define KEF_BOUND 0x0002 /* Thread can not migrate. */
#define KEF_XFERABLE 0x0004 /* Thread was added as transferable. */
#define KEF_HOLD 0x0008 /* Thread is temporarily bound. */
#define KEF_REMOVED 0x0010 /* Thread was removed while ASSIGNED */
Rework the interface between priority propagation (lending) and the schedulers a bit to ensure more correct handling of priorities and fewer priority inversions: - Add two functions to the sched(9) API to handle priority lending: sched_lend_prio() and sched_unlend_prio(). The turnstile code uses these functions to ask the scheduler to lend a thread a set priority and to tell the scheduler when it thinks it is ok for a thread to stop borrowing priority. The unlend case is slightly complex in that the turnstile code tells the scheduler what the minimum priority of the thread needs to be to satisfy the requirements of any other threads blocked on locks owned by the thread in question. The scheduler then decides where the thread can go back to normal mode (if it's normal priority is high enough to satisfy the pending lock requests) or it it should continue to use the priority specified to the sched_unlend_prio() call. This involves adding a new per-thread flag TDF_BORROWING that replaces the ULE-only kse flag for priority elevation. - Schedulers now refuse to lower the priority of a thread that is currently borrowing another therad's priority. - If a scheduler changes the priority of a thread that is currently sitting on a turnstile, it will call a new function turnstile_adjust() to inform the turnstile code of the change. This function resorts the thread on the priority list of the turnstile if needed, and if the thread ends up at the head of the list (due to having the highest priority) and its priority was raised, then it will propagate that new priority to the owner of the lock it is blocked on. Some additional fixes specific to the 4BSD scheduler include: - Common code for updating the priority of a thread when the user priority of its associated kse group has been consolidated in a new static function resetpriority_thread(). One change to this function is that it will now only adjust the priority of a thread if it already has a time sharing priority, thus preserving any boosts from a tsleep() until the thread returns to userland. Also, resetpriority() no longer calls maybe_resched() on each thread in the group. Instead, the code calling resetpriority() is responsible for calling resetpriority_thread() on any threads that need to be updated. - schedcpu() now uses resetpriority_thread() instead of just calling sched_prio() directly after it updates a kse group's user priority. - sched_clock() now uses resetpriority_thread() rather than writing directly to td_priority. - sched_nice() now updates all the priorities of the threads after the group priority has been adjusted. Discussed with: bde Reviewed by: ups, jeffr Tested on: 4bsd, ule Tested on: i386, alpha, sparc64
2004-12-30 20:52:44 +00:00
#define KEF_INTERNAL 0x0020
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
struct kg_sched {
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
struct thread *skg_last_assigned; /* (j) Last thread assigned to */
/* the system scheduler */
- Add a new variable 'kg_runtime' that tracks the amount of time we've run. - Use the ratio of kg_runtime / kg_slptime to determine our dynamic priority. - Scale kg_runtime and kg_slptime back when the sum of the two exceeds SCHED_SLP_RUN_MAX. This allows us to slowly forget old behavior. - Scale back the runtime and slptime in fork so that the new process has the same ratio but much less accumulated time. This causes new behavior to be noticed more quickly.
2003-02-10 14:03:45 +00:00
int skg_slptime; /* Number of ticks we vol. slept */
int skg_runtime; /* Number of ticks we were running */
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
int skg_avail_opennings; /* (j) Num unfilled slots in group.*/
int skg_concurrency; /* (j) Num threads requested in group.*/
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
};
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
#define kg_last_assigned kg_sched->skg_last_assigned
#define kg_avail_opennings kg_sched->skg_avail_opennings
#define kg_concurrency kg_sched->skg_concurrency
#define kg_runtime kg_sched->skg_runtime
#define kg_slptime kg_sched->skg_slptime
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
Use some macros to trach available scheduler slots to allow easier debugging. MFC after: 4 days
2004-10-05 21:10:44 +00:00
#define SLOT_RELEASE(kg) \
do { \
kg->kg_avail_opennings++; \
CTR3(KTR_RUNQ, "kg %p(%d) Slot released (->%d)", \
kg, \
kg->kg_concurrency, \
kg->kg_avail_opennings); \
/*KASSERT((kg->kg_avail_opennings <= kg->kg_concurrency), \
("slots out of whack")); */ \
} while (0)
#define SLOT_USE(kg) \
do { \
kg->kg_avail_opennings--; \
CTR3(KTR_RUNQ, "kg %p(%d) Slot used (->%d)", \
kg, \
kg->kg_concurrency, \
kg->kg_avail_opennings); \
/*KASSERT((kg->kg_avail_opennings >= 0), \
("slots out of whack"));*/ \
} while (0)
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
static struct kse kse0;
static struct kg_sched kg_sched0;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
/*
- Don't allow nice to have such a large effect on priority. This was causing poor interactive performance while unnice processes were running. The new scheme still allows nice to have an effect on priority but it is not as dramatic as the effect of the interactivity score.
2003-06-21 02:22:47 +00:00
* The priority is primarily determined by the interactivity score. Thus, we
* give lower(better) priorities to kse groups that use less CPU. The nice
* value is then directly added to this to allow nice to have some effect
* on latency.
- Create a function sched_interact_score() which decides on the interactivity of a kseg and assigns it a value of 0 through 100. - Use sched_interact_score() to determine the dynamic priority. - Define SCHED_CURR() in terms of sched_interact_score(). - Adjust the maximum slice back down to 100ms. - Remove redundant clearing of ke_runq in sched_wakeup() - Clean up #defines and comment them.
2003-03-04 02:45:59 +00:00
*
* PRI_RANGE: Total priority range for timeshare threads.
- Don't allow nice to have such a large effect on priority. This was causing poor interactive performance while unnice processes were running. The new scheme still allows nice to have an effect on priority but it is not as dramatic as the effect of the interactivity score.
2003-06-21 02:22:47 +00:00
* PRI_NRESV: Number of nice values.
- Create a function sched_interact_score() which decides on the interactivity of a kseg and assigns it a value of 0 through 100. - Use sched_interact_score() to determine the dynamic priority. - Define SCHED_CURR() in terms of sched_interact_score(). - Adjust the maximum slice back down to 100ms. - Remove redundant clearing of ke_runq in sched_wakeup() - Clean up #defines and comment them.
2003-03-04 02:45:59 +00:00
* PRI_BASE: The start of the dynamic range.
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
*/
- Create a function sched_interact_score() which decides on the interactivity of a kseg and assigns it a value of 0 through 100. - Use sched_interact_score() to determine the dynamic priority. - Define SCHED_CURR() in terms of sched_interact_score(). - Adjust the maximum slice back down to 100ms. - Remove redundant clearing of ke_runq in sched_wakeup() - Clean up #defines and comment them.
2003-03-04 02:45:59 +00:00
#define SCHED_PRI_RANGE (PRI_MAX_TIMESHARE - PRI_MIN_TIMESHARE + 1)
- Remove uses of PRIO_TOTAL and replace them with SCHED_PRI_NRESV - SCHED_PRI_NRESV does not have the off by one error in PRIO_TOTAL so we do not have to account for it in the few places that we use it. Requested by: bde
2003-11-02 03:49:32 +00:00
#define SCHED_PRI_NRESV ((PRIO_MAX - PRIO_MIN) + 1)
#define SCHED_PRI_NHALF (SCHED_PRI_NRESV / 2)
- Don't allow nice to have such a large effect on priority. This was causing poor interactive performance while unnice processes were running. The new scheme still allows nice to have an effect on priority but it is not as dramatic as the effect of the interactivity score.
2003-06-21 02:22:47 +00:00
#define SCHED_PRI_BASE (PRI_MIN_TIMESHARE)
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
#define SCHED_PRI_INTERACT(score) \
- Don't allow nice to have such a large effect on priority. This was causing poor interactive performance while unnice processes were running. The new scheme still allows nice to have an effect on priority but it is not as dramatic as the effect of the interactivity score.
2003-06-21 02:22:47 +00:00
((score) * SCHED_PRI_RANGE / SCHED_INTERACT_MAX)
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
/*
- Create a function sched_interact_score() which decides on the interactivity of a kseg and assigns it a value of 0 through 100. - Use sched_interact_score() to determine the dynamic priority. - Define SCHED_CURR() in terms of sched_interact_score(). - Adjust the maximum slice back down to 100ms. - Remove redundant clearing of ke_runq in sched_wakeup() - Clean up #defines and comment them.
2003-03-04 02:45:59 +00:00
* These determine the interactivity of a process.
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
*
- Add a new variable 'kg_runtime' that tracks the amount of time we've run. - Use the ratio of kg_runtime / kg_slptime to determine our dynamic priority. - Scale kg_runtime and kg_slptime back when the sum of the two exceeds SCHED_SLP_RUN_MAX. This allows us to slowly forget old behavior. - Scale back the runtime and slptime in fork so that the new process has the same ratio but much less accumulated time. This causes new behavior to be noticed more quickly.
2003-02-10 14:03:45 +00:00
* SLP_RUN_MAX: Maximum amount of sleep time + run time we'll accumulate
* before throttling back.
- Change sched_interact_update() to only accept slp+runtime values between 0 and SCHED_SLP_RUN_MAX * 2. This allows us to simplify the algorithm quite a bit. Before, it dealt with arbitrary values which required us to do nasty integer division tricks that didn't quite work out correctly. - Chnage sched_wakeup() to detect conditions where the slp+runtime could exceed SCHED_SLP_RUN_MAX * 2. This can happen if we go to sleep for longer than 6 seconds. In this case, we'll just clear the runtime and set the sleep time to the max. - Define a new function, sched_interact_fork() which updates the slp+runtime of a newly forked thread. We want to limit the amount of history retained from the parent so that we learn the child's behavior quickly. We don't, however want to decay it to nothing. Previously, we would simply divide each parameter by 100 whenever we forked. After a few forks the values would reach 0 and tasks would not be considered interactive. - Add another KTR entry, cleanup some existing entries. - Remove a useless sched_interact_update() from sched_priority(). This is already done by the callers that require it.
2003-11-02 03:36:33 +00:00
* SLP_RUN_FORK: Maximum slp+run time to inherit at fork time.
- Fix the maximum slice value. I accidentally checked in a value of '2' which meant no process would run for longer than 20ms. - Slightly redo the interactivity scorer. It follows the same algorithm but in a slightly more correct way. Previously values above half were incorrect. - Lower the interactivity threshold to 20. It seems that in testing non- interactive tasks are hardly ever near there and expensive interactive tasks can sometimes surpass it. This area needs more testing. - Remove an unnecessary KTR. - Fix a case where an idle thread that had an elevated priority due to priority prop. would be placed back on the idle queue. - Delay setting NEEDRESCHED until userret() for threads that haad their priority elevated while in kernel. This gives us the same context switch optimization as SCHED_4BSD. - Limit the child's slice to 1 in sched_fork_kse() so we detect its behavior more quickly. - Inhert some of the run/slp time from the child in sched_exit_ksegrp(). - Redo some of the priority comparisons so they are more clear. - Throttle the frequency of sched_pctcpu_update() so that rounding errors do not make it invalid.
2003-06-15 02:18:29 +00:00
* INTERACT_MAX: Maximum interactivity value. Smaller is better.
- Create a function sched_interact_score() which decides on the interactivity of a kseg and assigns it a value of 0 through 100. - Use sched_interact_score() to determine the dynamic priority. - Define SCHED_CURR() in terms of sched_interact_score(). - Adjust the maximum slice back down to 100ms. - Remove redundant clearing of ke_runq in sched_wakeup() - Clean up #defines and comment them.
2003-03-04 02:45:59 +00:00
* INTERACT_THRESH: Threshhold for placement on the current runq.
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
*/
- The non iterative algorithm for interact_update was broken due to rounding errors. This was the source of the majority of the interactivity problems. Reintroduce the old algorithm and its XXX. - Up the interactivity threshold to 30. It really could stand to be even a tiny bit higher. - Let the sleep and run time accumulate up to 5 seconds of history rather than two. This helps stop XFree86 from becoming non-interactive during bursts of activity.
2003-10-16 08:17:43 +00:00
#define SCHED_SLP_RUN_MAX ((hz * 5) << 10)
- Change sched_interact_update() to only accept slp+runtime values between 0 and SCHED_SLP_RUN_MAX * 2. This allows us to simplify the algorithm quite a bit. Before, it dealt with arbitrary values which required us to do nasty integer division tricks that didn't quite work out correctly. - Chnage sched_wakeup() to detect conditions where the slp+runtime could exceed SCHED_SLP_RUN_MAX * 2. This can happen if we go to sleep for longer than 6 seconds. In this case, we'll just clear the runtime and set the sleep time to the max. - Define a new function, sched_interact_fork() which updates the slp+runtime of a newly forked thread. We want to limit the amount of history retained from the parent so that we learn the child's behavior quickly. We don't, however want to decay it to nothing. Previously, we would simply divide each parameter by 100 whenever we forked. After a few forks the values would reach 0 and tasks would not be considered interactive. - Add another KTR entry, cleanup some existing entries. - Remove a useless sched_interact_update() from sched_priority(). This is already done by the callers that require it.
2003-11-02 03:36:33 +00:00
#define SCHED_SLP_RUN_FORK ((hz / 2) << 10)
- Fix the maximum slice value. I accidentally checked in a value of '2' which meant no process would run for longer than 20ms. - Slightly redo the interactivity scorer. It follows the same algorithm but in a slightly more correct way. Previously values above half were incorrect. - Lower the interactivity threshold to 20. It seems that in testing non- interactive tasks are hardly ever near there and expensive interactive tasks can sometimes surpass it. This area needs more testing. - Remove an unnecessary KTR. - Fix a case where an idle thread that had an elevated priority due to priority prop. would be placed back on the idle queue. - Delay setting NEEDRESCHED until userret() for threads that haad their priority elevated while in kernel. This gives us the same context switch optimization as SCHED_4BSD. - Limit the child's slice to 1 in sched_fork_kse() so we detect its behavior more quickly. - Inhert some of the run/slp time from the child in sched_exit_ksegrp(). - Redo some of the priority comparisons so they are more clear. - Throttle the frequency of sched_pctcpu_update() so that rounding errors do not make it invalid.
2003-06-15 02:18:29 +00:00
#define SCHED_INTERACT_MAX (100)
#define SCHED_INTERACT_HALF (SCHED_INTERACT_MAX / 2)
- The non iterative algorithm for interact_update was broken due to rounding errors. This was the source of the majority of the interactivity problems. Reintroduce the old algorithm and its XXX. - Up the interactivity threshold to 30. It really could stand to be even a tiny bit higher. - Let the sleep and run time accumulate up to 5 seconds of history rather than two. This helps stop XFree86 from becoming non-interactive during bursts of activity.
2003-10-16 08:17:43 +00:00
#define SCHED_INTERACT_THRESH (30)
- Create a function sched_interact_score() which decides on the interactivity of a kseg and assigns it a value of 0 through 100. - Use sched_interact_score() to determine the dynamic priority. - Define SCHED_CURR() in terms of sched_interact_score(). - Adjust the maximum slice back down to 100ms. - Remove redundant clearing of ke_runq in sched_wakeup() - Clean up #defines and comment them.
2003-03-04 02:45:59 +00:00
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
/*
* These parameters and macros determine the size of the time slice that is
* granted to each thread.
*
* SLICE_MIN: Minimum time slice granted, in units of ticks.
* SLICE_MAX: Maximum time slice granted.
* SLICE_RANGE: Range of available time slices scaled by hz.
- Add in support for KSEs with 0 slice values on the run queue. If we try to select a KSE with a slice of 0 we will update its slice and insert it onto the next queue. - Pass the KSE instead of the ksegrp into sched_slice(). This more accurately reflects the behavior of the code. Slices are granted to kses. - Add a function kseq_nice_min() which finds the smallest nice value assigned to the kseg of any KSE on the queue. - Rewrite the logic in sched_slice(). Add a large comment describing the new slice selection scheme. To summarize, slices are assigned based on the nice value. Priorities are still calculated based on the nice and interactivity of a process. Slice sizes of 0 may be granted for KSEs whos nice is 20 or futher away from the lowest nice on the run queue. Other nice values are scaled across the range [min, min+20]. This fixes ULEs bad behavior with positively niced processes.
2003-04-02 06:46:43 +00:00
* SLICE_SCALE: The number slices granted per val in the range of [0, max].
* SLICE_NICE: Determine the amount of slice granted to a scaled nice.
- Rename SCHED_PRI_NTHRESH to SCHED_SLICE_NTHRESH since it is only used in slice assignment. Add a comment describing what it does. - Remove a stale XXX comment, the nice should not impact the interactivity, nice adjustments only effect non-interactive tasks in ULE. - Don't allow nice -20 tasks to totally starve nice 0 tasks. Give them at least SCHED_SLICE_MIN ticks. We still allow nice 0 tasks to starve nice +20 tasks as intended.
2003-11-02 04:10:15 +00:00
* SLICE_NTHRESH: The nice cutoff point for slice assignment.
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
*/
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
#define SCHED_SLICE_MIN (slice_min)
#define SCHED_SLICE_MAX (slice_max)
- Allow interactive tasks to use the maximum time-slice. This is not as detrimental as I thought it would be in the case of massive process storms from a shell and it makes regular desktop usage noticeably better.
2004-02-01 10:38:13 +00:00
#define SCHED_SLICE_INTERACTIVE (slice_max)
- Rename SCHED_PRI_NTHRESH to SCHED_SLICE_NTHRESH since it is only used in slice assignment. Add a comment describing what it does. - Remove a stale XXX comment, the nice should not impact the interactivity, nice adjustments only effect non-interactive tasks in ULE. - Don't allow nice -20 tasks to totally starve nice 0 tasks. Give them at least SCHED_SLICE_MIN ticks. We still allow nice 0 tasks to starve nice +20 tasks as intended.
2003-11-02 04:10:15 +00:00
#define SCHED_SLICE_NTHRESH (SCHED_PRI_NHALF - 1)
- Create a function sched_interact_score() which decides on the interactivity of a kseg and assigns it a value of 0 through 100. - Use sched_interact_score() to determine the dynamic priority. - Define SCHED_CURR() in terms of sched_interact_score(). - Adjust the maximum slice back down to 100ms. - Remove redundant clearing of ke_runq in sched_wakeup() - Clean up #defines and comment them.
2003-03-04 02:45:59 +00:00
#define SCHED_SLICE_RANGE (SCHED_SLICE_MAX - SCHED_SLICE_MIN + 1)
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
#define SCHED_SLICE_SCALE(val, max) (((val) * SCHED_SLICE_RANGE) / (max))
- Add in support for KSEs with 0 slice values on the run queue. If we try to select a KSE with a slice of 0 we will update its slice and insert it onto the next queue. - Pass the KSE instead of the ksegrp into sched_slice(). This more accurately reflects the behavior of the code. Slices are granted to kses. - Add a function kseq_nice_min() which finds the smallest nice value assigned to the kseg of any KSE on the queue. - Rewrite the logic in sched_slice(). Add a large comment describing the new slice selection scheme. To summarize, slices are assigned based on the nice value. Priorities are still calculated based on the nice and interactivity of a process. Slice sizes of 0 may be granted for KSEs whos nice is 20 or futher away from the lowest nice on the run queue. Other nice values are scaled across the range [min, min+20]. This fixes ULEs bad behavior with positively niced processes.
2003-04-02 06:46:43 +00:00
#define SCHED_SLICE_NICE(nice) \
- Rename SCHED_PRI_NTHRESH to SCHED_SLICE_NTHRESH since it is only used in slice assignment. Add a comment describing what it does. - Remove a stale XXX comment, the nice should not impact the interactivity, nice adjustments only effect non-interactive tasks in ULE. - Don't allow nice -20 tasks to totally starve nice 0 tasks. Give them at least SCHED_SLICE_MIN ticks. We still allow nice 0 tasks to starve nice +20 tasks as intended.
2003-11-02 04:10:15 +00:00
(SCHED_SLICE_MAX - SCHED_SLICE_SCALE((nice), SCHED_SLICE_NTHRESH))
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
/*
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
* This macro determines whether or not the thread belongs on the current or
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
* next run queue.
*/
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
#define SCHED_INTERACTIVE(kg) \
(sched_interact_score(kg) < SCHED_INTERACT_THRESH)
- Unbreak priority prop. for timeshare threads. Always place something on the current queue if its priority is really elevated. This needs more work as there are cases where a next queue kse could be holding up what would be a curr queue kse, and thus hurting interactivity. Also, when a thread with an elevated priority has its priority lowered it should be placed back on the next queue.
2003-04-12 22:33:24 +00:00
#define SCHED_CURR(kg, ke) \
Rework the interface between priority propagation (lending) and the schedulers a bit to ensure more correct handling of priorities and fewer priority inversions: - Add two functions to the sched(9) API to handle priority lending: sched_lend_prio() and sched_unlend_prio(). The turnstile code uses these functions to ask the scheduler to lend a thread a set priority and to tell the scheduler when it thinks it is ok for a thread to stop borrowing priority. The unlend case is slightly complex in that the turnstile code tells the scheduler what the minimum priority of the thread needs to be to satisfy the requirements of any other threads blocked on locks owned by the thread in question. The scheduler then decides where the thread can go back to normal mode (if it's normal priority is high enough to satisfy the pending lock requests) or it it should continue to use the priority specified to the sched_unlend_prio() call. This involves adding a new per-thread flag TDF_BORROWING that replaces the ULE-only kse flag for priority elevation. - Schedulers now refuse to lower the priority of a thread that is currently borrowing another therad's priority. - If a scheduler changes the priority of a thread that is currently sitting on a turnstile, it will call a new function turnstile_adjust() to inform the turnstile code of the change. This function resorts the thread on the priority list of the turnstile if needed, and if the thread ends up at the head of the list (due to having the highest priority) and its priority was raised, then it will propagate that new priority to the owner of the lock it is blocked on. Some additional fixes specific to the 4BSD scheduler include: - Common code for updating the priority of a thread when the user priority of its associated kse group has been consolidated in a new static function resetpriority_thread(). One change to this function is that it will now only adjust the priority of a thread if it already has a time sharing priority, thus preserving any boosts from a tsleep() until the thread returns to userland. Also, resetpriority() no longer calls maybe_resched() on each thread in the group. Instead, the code calling resetpriority() is responsible for calling resetpriority_thread() on any threads that need to be updated. - schedcpu() now uses resetpriority_thread() instead of just calling sched_prio() directly after it updates a kse group's user priority. - sched_clock() now uses resetpriority_thread() rather than writing directly to td_priority. - sched_nice() now updates all the priorities of the threads after the group priority has been adjusted. Discussed with: bde Reviewed by: ups, jeffr Tested on: 4bsd, ule Tested on: i386, alpha, sparc64
2004-12-30 20:52:44 +00:00
((ke->ke_thread->td_flags & TDF_BORROWING) || SCHED_INTERACTIVE(kg))
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
/*
* Cpu percentage computation macros and defines.
*
* SCHED_CPU_TIME: Number of seconds to average the cpu usage across.
* SCHED_CPU_TICKS: Number of hz ticks to average the cpu usage across.
*/
- Make the interactivity calculator decay faster. - Make the pcpu estimator update faster.
2003-04-02 08:22:33 +00:00
#define SCHED_CPU_TIME 10
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
#define SCHED_CPU_TICKS (hz * SCHED_CPU_TIME)
/*
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
* kseq - per processor runqs and statistics.
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
*/
struct kseq {
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
struct runq ksq_idle; /* Queue of IDLE threads. */
struct runq ksq_timeshare[2]; /* Run queues for !IDLE. */
struct runq *ksq_next; /* Next timeshare queue. */
struct runq *ksq_curr; /* Current queue. */
- Remove the ksq_loads[] array. We are only interested in three counts, the total load, the timeshare load, and the number of threads that can be migrated to another cpu. Account for these seperately. - Introduce a KSE_CAN_MIGRATE() macro which determines whether or not a KSE can be migrated to another CPU. Currently, this only checks to see if we're an interrupt handler. Eventually this will also be used to support CPU binding.
2003-11-02 10:56:48 +00:00
int ksq_load_timeshare; /* Load for timeshare. */
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
int ksq_load; /* Aggregate load. */
- Remove uses of PRIO_TOTAL and replace them with SCHED_PRI_NRESV - SCHED_PRI_NRESV does not have the off by one error in PRIO_TOTAL so we do not have to account for it in the few places that we use it. Requested by: bde
2003-11-02 03:49:32 +00:00
short ksq_nice[SCHED_PRI_NRESV]; /* KSEs in each nice bin. */
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
short ksq_nicemin; /* Least nice. */
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
#ifdef SMP
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
int ksq_transferable;
LIST_ENTRY(kseq) ksq_siblings; /* Next in kseq group. */
struct kseq_group *ksq_group; /* Our processor group. */
volatile struct kse *ksq_assigned; /* assigned by another CPU. */
- Add a new member to struct kseq called ksq_sysload. This is intended to track the load for the sched_load() function. In the SMP case this member is not defined because it would be redundant with the ksg_load member which already tracks the non ithd load. - For sched_load() in the UP case simply return ksq_sysload. In the SMP case traverse the list of kseq groups and sum up their ksg_load fields.
2004-02-01 02:48:36 +00:00
#else
int ksq_sysload; /* For loadavg, !ITHD load. */
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
#endif
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
};
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
#ifdef SMP
/*
* kseq groups are groups of processors which can cheaply share threads. When
* one processor in the group goes idle it will check the runqs of the other
* processors in its group prior to halting and waiting for an interrupt.
* These groups are suitable for SMT (Symetric Multi-Threading) and not NUMA.
* In a numa environment we'd want an idle bitmap per group and a two tiered
* load balancer.
*/
struct kseq_group {
int ksg_cpus; /* Count of CPUs in this kseq group. */
Change the type of the various CPU masks to cpumask_t. Note that as long as there are still explicit uses of int, whether in types or in function names (such as atomic_set_int() in sched_ule.c), we can not change cpumask_t to be anything other than u_int. See also the commit log for sys/sys/types.h, revision 1.84.
2004-03-27 18:21:24 +00:00
cpumask_t ksg_cpumask; /* Mask of cpus in this group. */
cpumask_t ksg_idlemask; /* Idle cpus in this group. */
cpumask_t ksg_mask; /* Bit mask for first cpu. */
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
int ksg_load; /* Total load of this group. */
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
int ksg_transferable; /* Transferable load of this group. */
LIST_HEAD(, kseq) ksg_members; /* Linked list of all members. */
};
#endif
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
/*
* One kse queue per processor.
*/
- Use ksq_load as the authoritive count of kses on the pair of kseqs for sched_runnable() et all. - Remove some dead code in sched_clock(). - Define two macros KSEQ_SELF() and KSEQ_CPU() for getting the kseq of the current cpu or some alternate cpu. - Start introducing kseq_() functions, such as kseq_choose() and kseq_setup().
2003-01-29 07:00:51 +00:00
#ifdef SMP
Change the type of the various CPU masks to cpumask_t. Note that as long as there are still explicit uses of int, whether in types or in function names (such as atomic_set_int() in sched_ule.c), we can not change cpumask_t to be anything other than u_int. See also the commit log for sys/sys/types.h, revision 1.84.
2004-03-27 18:21:24 +00:00
static cpumask_t kseq_idle;
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
static int ksg_maxid;
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
static struct kseq kseq_cpu[MAXCPU];
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
static struct kseq_group kseq_groups[MAXCPU];
- Run sched_balance() and sched_balance_groups() from hardclock via sched_clock() rather than using callouts. This means we no longer have to take the load of the callout thread into consideration while balancing and should make the balancing decisions simpler and more accurate. Tested on: x86/UP, amd64/SMP
2004-06-02 05:46:48 +00:00
static int bal_tick;
static int gbal_tick;
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
static int balance_groups;
- Run sched_balance() and sched_balance_groups() from hardclock via sched_clock() rather than using callouts. This means we no longer have to take the load of the callout thread into consideration while balancing and should make the balancing decisions simpler and more accurate. Tested on: x86/UP, amd64/SMP
2004-06-02 05:46:48 +00:00
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
#define KSEQ_SELF() (&kseq_cpu[PCPU_GET(cpuid)])
#define KSEQ_CPU(x) (&kseq_cpu[(x)])
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
#define KSEQ_ID(x) ((x) - kseq_cpu)
#define KSEQ_GROUP(x) (&kseq_groups[(x)])
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
#else /* !SMP */
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
static struct kseq kseq_cpu;
- Run sched_balance() and sched_balance_groups() from hardclock via sched_clock() rather than using callouts. This means we no longer have to take the load of the callout thread into consideration while balancing and should make the balancing decisions simpler and more accurate. Tested on: x86/UP, amd64/SMP
2004-06-02 05:46:48 +00:00
- Use ksq_load as the authoritive count of kses on the pair of kseqs for sched_runnable() et all. - Remove some dead code in sched_clock(). - Define two macros KSEQ_SELF() and KSEQ_CPU() for getting the kseq of the current cpu or some alternate cpu. - Start introducing kseq_() functions, such as kseq_choose() and kseq_setup().
2003-01-29 07:00:51 +00:00
#define KSEQ_SELF() (&kseq_cpu)
#define KSEQ_CPU(x) (&kseq_cpu)
#endif
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
static void slot_fill(struct ksegrp *kg);
static struct kse *sched_choose(void); /* XXX Should be thread * */
- Add in support for KSEs with 0 slice values on the run queue. If we try to select a KSE with a slice of 0 we will update its slice and insert it onto the next queue. - Pass the KSE instead of the ksegrp into sched_slice(). This more accurately reflects the behavior of the code. Slices are granted to kses. - Add a function kseq_nice_min() which finds the smallest nice value assigned to the kseg of any KSE on the queue. - Rewrite the logic in sched_slice(). Add a large comment describing the new slice selection scheme. To summarize, slices are assigned based on the nice value. Priorities are still calculated based on the nice and interactivity of a process. Slice sizes of 0 may be granted for KSEs whos nice is 20 or futher away from the lowest nice on the run queue. Other nice values are scaled across the range [min, min+20]. This fixes ULEs bad behavior with positively niced processes.
2003-04-02 06:46:43 +00:00
static void sched_slice(struct kse *ke);
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
static void sched_priority(struct ksegrp *kg);
Rework the interface between priority propagation (lending) and the schedulers a bit to ensure more correct handling of priorities and fewer priority inversions: - Add two functions to the sched(9) API to handle priority lending: sched_lend_prio() and sched_unlend_prio(). The turnstile code uses these functions to ask the scheduler to lend a thread a set priority and to tell the scheduler when it thinks it is ok for a thread to stop borrowing priority. The unlend case is slightly complex in that the turnstile code tells the scheduler what the minimum priority of the thread needs to be to satisfy the requirements of any other threads blocked on locks owned by the thread in question. The scheduler then decides where the thread can go back to normal mode (if it's normal priority is high enough to satisfy the pending lock requests) or it it should continue to use the priority specified to the sched_unlend_prio() call. This involves adding a new per-thread flag TDF_BORROWING that replaces the ULE-only kse flag for priority elevation. - Schedulers now refuse to lower the priority of a thread that is currently borrowing another therad's priority. - If a scheduler changes the priority of a thread that is currently sitting on a turnstile, it will call a new function turnstile_adjust() to inform the turnstile code of the change. This function resorts the thread on the priority list of the turnstile if needed, and if the thread ends up at the head of the list (due to having the highest priority) and its priority was raised, then it will propagate that new priority to the owner of the lock it is blocked on. Some additional fixes specific to the 4BSD scheduler include: - Common code for updating the priority of a thread when the user priority of its associated kse group has been consolidated in a new static function resetpriority_thread(). One change to this function is that it will now only adjust the priority of a thread if it already has a time sharing priority, thus preserving any boosts from a tsleep() until the thread returns to userland. Also, resetpriority() no longer calls maybe_resched() on each thread in the group. Instead, the code calling resetpriority() is responsible for calling resetpriority_thread() on any threads that need to be updated. - schedcpu() now uses resetpriority_thread() instead of just calling sched_prio() directly after it updates a kse group's user priority. - sched_clock() now uses resetpriority_thread() rather than writing directly to td_priority. - sched_nice() now updates all the priorities of the threads after the group priority has been adjusted. Discussed with: bde Reviewed by: ups, jeffr Tested on: 4bsd, ule Tested on: i386, alpha, sparc64
2004-12-30 20:52:44 +00:00
static void sched_thread_priority(struct thread *td, u_char prio);
- Create a function sched_interact_score() which decides on the interactivity of a kseg and assigns it a value of 0 through 100. - Use sched_interact_score() to determine the dynamic priority. - Define SCHED_CURR() in terms of sched_interact_score(). - Adjust the maximum slice back down to 100ms. - Remove redundant clearing of ke_runq in sched_wakeup() - Clean up #defines and comment them.
2003-03-04 02:45:59 +00:00
static int sched_interact_score(struct ksegrp *kg);
- Add a new function "sched_interact_update()" that scales back the sleep and run time. - Scale the sleep and run time back via sched_interact_update() in more places. This is to keep the statistic more accurate. - Charge a parent one tick for forking a child. - Add only the run time and not the sleep time to the parents kg when a thread exits. This allows us to give a penalty for having an expensive thread exit but does not give a bonus for having an interactive thread exit. - Change the SLP_RUN_THROTTLE to limit us to 4/5th and not 1/2. - Change the SLP_RUN_MAX to two seconds. This keeps bursty interactive applications like mozilla and openoffice in the interactive range even through expensive tasks. - Recalculate the slice after every sleep. This ensures that once a task has been marked interactive it only has a slice of 1 at the risk of giving tasks that sleep for a very brief period a longer time slice.
2003-06-17 06:39:51 +00:00
static void sched_interact_update(struct ksegrp *kg);
- Change sched_interact_update() to only accept slp+runtime values between 0 and SCHED_SLP_RUN_MAX * 2. This allows us to simplify the algorithm quite a bit. Before, it dealt with arbitrary values which required us to do nasty integer division tricks that didn't quite work out correctly. - Chnage sched_wakeup() to detect conditions where the slp+runtime could exceed SCHED_SLP_RUN_MAX * 2. This can happen if we go to sleep for longer than 6 seconds. In this case, we'll just clear the runtime and set the sleep time to the max. - Define a new function, sched_interact_fork() which updates the slp+runtime of a newly forked thread. We want to limit the amount of history retained from the parent so that we learn the child's behavior quickly. We don't, however want to decay it to nothing. Previously, we would simply divide each parameter by 100 whenever we forked. After a few forks the values would reach 0 and tasks would not be considered interactive. - Add another KTR entry, cleanup some existing entries. - Remove a useless sched_interact_update() from sched_priority(). This is already done by the callers that require it.
2003-11-02 03:36:33 +00:00
static void sched_interact_fork(struct ksegrp *kg);
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
static void sched_pctcpu_update(struct kse *ke);
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
/* Operations on per processor queues */
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
static struct kse * kseq_choose(struct kseq *kseq);
- Use ksq_load as the authoritive count of kses on the pair of kseqs for sched_runnable() et all. - Remove some dead code in sched_clock(). - Define two macros KSEQ_SELF() and KSEQ_CPU() for getting the kseq of the current cpu or some alternate cpu. - Start introducing kseq_() functions, such as kseq_choose() and kseq_setup().
2003-01-29 07:00:51 +00:00
static void kseq_setup(struct kseq *kseq);
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
static void kseq_load_add(struct kseq *kseq, struct kse *ke);
static void kseq_load_rem(struct kseq *kseq, struct kse *ke);
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
static __inline void kseq_runq_add(struct kseq *kseq, struct kse *ke, int);
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
static __inline void kseq_runq_rem(struct kseq *kseq, struct kse *ke);
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
static void kseq_nice_add(struct kseq *kseq, int nice);
static void kseq_nice_rem(struct kseq *kseq, int nice);
- Set the ke_cpu field in sched_add() for interrupt and realtime threads since they are going on the current cpu and not their previously assigned cpu. - sched_runnable() should only return true in the SMP case if the other processor has more than one thread that is runnable. We can not steal curthread. - Change kseq_print() to accept the cpuid instead of a kseq pointer. This makes use of this function in ddb much easier.
2003-04-18 05:24:10 +00:00
void kseq_print(int cpu);
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
#ifdef SMP
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
static int kseq_transfer(struct kseq *ksq, struct kse *ke, int class);
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
static struct kse *runq_steal(struct runq *rq);
- Run sched_balance() and sched_balance_groups() from hardclock via sched_clock() rather than using callouts. This means we no longer have to take the load of the callout thread into consideration while balancing and should make the balancing decisions simpler and more accurate. Tested on: x86/UP, amd64/SMP
2004-06-02 05:46:48 +00:00
static void sched_balance(void);
static void sched_balance_groups(void);
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
static void sched_balance_group(struct kseq_group *ksg);
static void sched_balance_pair(struct kseq *high, struct kseq *low);
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
static void kseq_move(struct kseq *from, int cpu);
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
static int kseq_idled(struct kseq *kseq);
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
static void kseq_notify(struct kse *ke, int cpu);
static void kseq_assign(struct kseq *);
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
static struct kse *kseq_steal(struct kseq *kseq, int stealidle);
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
#define KSE_CAN_MIGRATE(ke) \
Revert the previous round of changes to td_pinned. The scheduler isn't fully initialed when the pmap layer tries to call sched_pini() early in the boot and results in an quick panic. Use ke_pinned instead as was originally done with Tor's patch. Approved by: julian
2004-09-11 10:07:22 +00:00
((ke)->ke_thread->td_pinned == 0 && ((ke)->ke_flags & KEF_BOUND) == 0)
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
#endif
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
void
- Set the ke_cpu field in sched_add() for interrupt and realtime threads since they are going on the current cpu and not their previously assigned cpu. - sched_runnable() should only return true in the SMP case if the other processor has more than one thread that is runnable. We can not steal curthread. - Change kseq_print() to accept the cpuid instead of a kseq pointer. This makes use of this function in ddb much easier.
2003-04-18 05:24:10 +00:00
kseq_print(int cpu)
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
{
- Set the ke_cpu field in sched_add() for interrupt and realtime threads since they are going on the current cpu and not their previously assigned cpu. - sched_runnable() should only return true in the SMP case if the other processor has more than one thread that is runnable. We can not steal curthread. - Change kseq_print() to accept the cpuid instead of a kseq pointer. This makes use of this function in ddb much easier.
2003-04-18 05:24:10 +00:00
struct kseq *kseq;
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
int i;
- Keep seperate statistics and run queues for different scheduling classes. - Treat each class specially in kseq_{choose,add,rem}. Let the rest of the code be less aware of scheduling classes. - Skip the interactivity calculation for non TIMESHARE ksegrps. - Move slice and runq selection into kseq_add(). Uninline it now that it's big.
2003-04-03 00:29:28 +00:00
- Set the ke_cpu field in sched_add() for interrupt and realtime threads since they are going on the current cpu and not their previously assigned cpu. - sched_runnable() should only return true in the SMP case if the other processor has more than one thread that is runnable. We can not steal curthread. - Change kseq_print() to accept the cpuid instead of a kseq pointer. This makes use of this function in ddb much easier.
2003-04-18 05:24:10 +00:00
kseq = KSEQ_CPU(cpu);
- Keep seperate statistics and run queues for different scheduling classes. - Treat each class specially in kseq_{choose,add,rem}. Let the rest of the code be less aware of scheduling classes. - Skip the interactivity calculation for non TIMESHARE ksegrps. - Move slice and runq selection into kseq_add(). Uninline it now that it's big.
2003-04-03 00:29:28 +00:00
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
printf("kseq:\n");
printf("\tload: %d\n", kseq->ksq_load);
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
printf("\tload TIMESHARE: %d\n", kseq->ksq_load_timeshare);
- Remove the ksq_loads[] array. We are only interested in three counts, the total load, the timeshare load, and the number of threads that can be migrated to another cpu. Account for these seperately. - Introduce a KSE_CAN_MIGRATE() macro which determines whether or not a KSE can be migrated to another CPU. Currently, this only checks to see if we're an interrupt handler. Eventually this will also be used to support CPU binding.
2003-11-02 10:56:48 +00:00
#ifdef SMP
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
printf("\tload transferable: %d\n", kseq->ksq_transferable);
- Remove the ksq_loads[] array. We are only interested in three counts, the total load, the timeshare load, and the number of threads that can be migrated to another cpu. Account for these seperately. - Introduce a KSE_CAN_MIGRATE() macro which determines whether or not a KSE can be migrated to another CPU. Currently, this only checks to see if we're an interrupt handler. Eventually this will also be used to support CPU binding.
2003-11-02 10:56:48 +00:00
#endif
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
printf("\tnicemin:\t%d\n", kseq->ksq_nicemin);
printf("\tnice counts:\n");
- Remove uses of PRIO_TOTAL and replace them with SCHED_PRI_NRESV - SCHED_PRI_NRESV does not have the off by one error in PRIO_TOTAL so we do not have to account for it in the few places that we use it. Requested by: bde
2003-11-02 03:49:32 +00:00
for (i = 0; i < SCHED_PRI_NRESV; i++)
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
if (kseq->ksq_nice[i])
printf("\t\t%d = %d\n",
i - SCHED_PRI_NHALF, kseq->ksq_nice[i]);
}
- Keep seperate statistics and run queues for different scheduling classes. - Treat each class specially in kseq_{choose,add,rem}. Let the rest of the code be less aware of scheduling classes. - Skip the interactivity calculation for non TIMESHARE ksegrps. - Move slice and runq selection into kseq_add(). Uninline it now that it's big.
2003-04-03 00:29:28 +00:00
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
static __inline void
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
kseq_runq_add(struct kseq *kseq, struct kse *ke, int flags)
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
{
#ifdef SMP
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
if (KSE_CAN_MIGRATE(ke)) {
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
kseq->ksq_transferable++;
kseq->ksq_group->ksg_transferable++;
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
ke->ke_flags |= KEF_XFERABLE;
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
}
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
#endif
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
runq_add(ke->ke_runq, ke, flags);
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
}
static __inline void
kseq_runq_rem(struct kseq *kseq, struct kse *ke)
{
#ifdef SMP
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
if (ke->ke_flags & KEF_XFERABLE) {
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
kseq->ksq_transferable--;
kseq->ksq_group->ksg_transferable--;
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
ke->ke_flags &= ~KEF_XFERABLE;
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
}
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
#endif
runq_remove(ke->ke_runq, ke);
}
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
static void
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
kseq_load_add(struct kseq *kseq, struct kse *ke)
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
{
- Remove the ksq_loads[] array. We are only interested in three counts, the total load, the timeshare load, and the number of threads that can be migrated to another cpu. Account for these seperately. - Introduce a KSE_CAN_MIGRATE() macro which determines whether or not a KSE can be migrated to another CPU. Currently, this only checks to see if we're an interrupt handler. Eventually this will also be used to support CPU binding.
2003-11-02 10:56:48 +00:00
int class;
- When a new thread is added to a kseq the load is incremented prior to adding it to the nice tables. Therefore, in kseq_add_nice, we should keep in mind that the load will be 1 if we are the only thread, and not 0. - Assert that the sched lock is held in all the appropriate places. - Increase the scope of the sched lock in sched_pctcpu_update(). - Hold the sched lock in sched_runnable(). It is not held by the caller.
2003-06-08 00:47:33 +00:00
mtx_assert(&sched_lock, MA_OWNED);
- Remove the ksq_loads[] array. We are only interested in three counts, the total load, the timeshare load, and the number of threads that can be migrated to another cpu. Account for these seperately. - Introduce a KSE_CAN_MIGRATE() macro which determines whether or not a KSE can be migrated to another CPU. Currently, this only checks to see if we're an interrupt handler. Eventually this will also be used to support CPU binding.
2003-11-02 10:56:48 +00:00
class = PRI_BASE(ke->ke_ksegrp->kg_pri_class);
if (class == PRI_TIMESHARE)
kseq->ksq_load_timeshare++;
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
kseq->ksq_load++;
- Remove earlier KTR_ULE tracepoints. - Define new KTR_SCHED points so that we can graph the operation of the scheduler.
2004-12-26 00:15:33 +00:00
CTR1(KTR_SCHED, "load: %d", kseq->ksq_load);
There was a thread on "unusually high load averages" when running under sched_ule, in January 2004. Looking at this, "pagezero" is (one of) the culprit(s). We had no provision for processes with P_NOLOAD set. With pagezero not running at PRI_ITHD, kseq_load_{add,rem} count pagezero as another-normal-process, thus the "expected-plus-one" load reported in the above thread. Submitted by: Nikos Ntarmos <ntarmos@ceid.upatras.gr>
2004-04-22 21:37:46 +00:00
if (class != PRI_ITHD && (ke->ke_proc->p_flag & P_NOLOAD) == 0)
- Add a new member to struct kseq called ksq_sysload. This is intended to track the load for the sched_load() function. In the SMP case this member is not defined because it would be redundant with the ksg_load member which already tracks the non ithd load. - For sched_load() in the UP case simply return ksq_sysload. In the SMP case traverse the list of kseq groups and sum up their ksg_load fields.
2004-02-01 02:48:36 +00:00
#ifdef SMP
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
kseq->ksq_group->ksg_load++;
- Add a new member to struct kseq called ksq_sysload. This is intended to track the load for the sched_load() function. In the SMP case this member is not defined because it would be redundant with the ksg_load member which already tracks the non ithd load. - For sched_load() in the UP case simply return ksq_sysload. In the SMP case traverse the list of kseq groups and sum up their ksg_load fields.
2004-02-01 02:48:36 +00:00
#else
kseq->ksq_sysload++;
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
#endif
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
if (ke->ke_ksegrp->kg_pri_class == PRI_TIMESHARE)
Nice, is a property of a process as a whole.. I mistakenly moved it to the ksegroup when breaking up the process structure. Put it back in the proc structure.
2004-06-16 00:26:31 +00:00
kseq_nice_add(kseq, ke->ke_proc->p_nice);
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
}
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
- Keep seperate statistics and run queues for different scheduling classes. - Treat each class specially in kseq_{choose,add,rem}. Let the rest of the code be less aware of scheduling classes. - Skip the interactivity calculation for non TIMESHARE ksegrps. - Move slice and runq selection into kseq_add(). Uninline it now that it's big.
2003-04-03 00:29:28 +00:00
static void
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
kseq_load_rem(struct kseq *kseq, struct kse *ke)
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
{
- Remove the ksq_loads[] array. We are only interested in three counts, the total load, the timeshare load, and the number of threads that can be migrated to another cpu. Account for these seperately. - Introduce a KSE_CAN_MIGRATE() macro which determines whether or not a KSE can be migrated to another CPU. Currently, this only checks to see if we're an interrupt handler. Eventually this will also be used to support CPU binding.
2003-11-02 10:56:48 +00:00
int class;
- When a new thread is added to a kseq the load is incremented prior to adding it to the nice tables. Therefore, in kseq_add_nice, we should keep in mind that the load will be 1 if we are the only thread, and not 0. - Assert that the sched lock is held in all the appropriate places. - Increase the scope of the sched lock in sched_pctcpu_update(). - Hold the sched lock in sched_runnable(). It is not held by the caller.
2003-06-08 00:47:33 +00:00
mtx_assert(&sched_lock, MA_OWNED);
- Remove the ksq_loads[] array. We are only interested in three counts, the total load, the timeshare load, and the number of threads that can be migrated to another cpu. Account for these seperately. - Introduce a KSE_CAN_MIGRATE() macro which determines whether or not a KSE can be migrated to another CPU. Currently, this only checks to see if we're an interrupt handler. Eventually this will also be used to support CPU binding.
2003-11-02 10:56:48 +00:00
class = PRI_BASE(ke->ke_ksegrp->kg_pri_class);
if (class == PRI_TIMESHARE)
kseq->ksq_load_timeshare--;
There was a thread on "unusually high load averages" when running under sched_ule, in January 2004. Looking at this, "pagezero" is (one of) the culprit(s). We had no provision for processes with P_NOLOAD set. With pagezero not running at PRI_ITHD, kseq_load_{add,rem} count pagezero as another-normal-process, thus the "expected-plus-one" load reported in the above thread. Submitted by: Nikos Ntarmos <ntarmos@ceid.upatras.gr>
2004-04-22 21:37:46 +00:00
if (class != PRI_ITHD && (ke->ke_proc->p_flag & P_NOLOAD) == 0)
- Add a new member to struct kseq called ksq_sysload. This is intended to track the load for the sched_load() function. In the SMP case this member is not defined because it would be redundant with the ksg_load member which already tracks the non ithd load. - For sched_load() in the UP case simply return ksq_sysload. In the SMP case traverse the list of kseq groups and sum up their ksg_load fields.
2004-02-01 02:48:36 +00:00
#ifdef SMP
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
kseq->ksq_group->ksg_load--;
- Add a new member to struct kseq called ksq_sysload. This is intended to track the load for the sched_load() function. In the SMP case this member is not defined because it would be redundant with the ksg_load member which already tracks the non ithd load. - For sched_load() in the UP case simply return ksq_sysload. In the SMP case traverse the list of kseq groups and sum up their ksg_load fields.
2004-02-01 02:48:36 +00:00
#else
kseq->ksq_sysload--;
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
#endif
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
kseq->ksq_load--;
- Remove earlier KTR_ULE tracepoints. - Define new KTR_SCHED points so that we can graph the operation of the scheduler.
2004-12-26 00:15:33 +00:00
CTR1(KTR_SCHED, "load: %d", kseq->ksq_load);
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
ke->ke_runq = NULL;
if (ke->ke_ksegrp->kg_pri_class == PRI_TIMESHARE)
Nice, is a property of a process as a whole.. I mistakenly moved it to the ksegroup when breaking up the process structure. Put it back in the proc structure.
2004-06-16 00:26:31 +00:00
kseq_nice_rem(kseq, ke->ke_proc->p_nice);
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
}
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
static void
kseq_nice_add(struct kseq *kseq, int nice)
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
{
- When a new thread is added to a kseq the load is incremented prior to adding it to the nice tables. Therefore, in kseq_add_nice, we should keep in mind that the load will be 1 if we are the only thread, and not 0. - Assert that the sched lock is held in all the appropriate places. - Increase the scope of the sched lock in sched_pctcpu_update(). - Hold the sched lock in sched_runnable(). It is not held by the caller.
2003-06-08 00:47:33 +00:00
mtx_assert(&sched_lock, MA_OWNED);
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
/* Normalize to zero. */
kseq->ksq_nice[nice + SCHED_PRI_NHALF]++;
- Remove the ksq_loads[] array. We are only interested in three counts, the total load, the timeshare load, and the number of threads that can be migrated to another cpu. Account for these seperately. - Introduce a KSE_CAN_MIGRATE() macro which determines whether or not a KSE can be migrated to another CPU. Currently, this only checks to see if we're an interrupt handler. Eventually this will also be used to support CPU binding.
2003-11-02 10:56:48 +00:00
if (nice < kseq->ksq_nicemin || kseq->ksq_load_timeshare == 1)
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
kseq->ksq_nicemin = nice;
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
}
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
static void
kseq_nice_rem(struct kseq *kseq, int nice)
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
{
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
int n;
- When a new thread is added to a kseq the load is incremented prior to adding it to the nice tables. Therefore, in kseq_add_nice, we should keep in mind that the load will be 1 if we are the only thread, and not 0. - Assert that the sched lock is held in all the appropriate places. - Increase the scope of the sched lock in sched_pctcpu_update(). - Hold the sched lock in sched_runnable(). It is not held by the caller.
2003-06-08 00:47:33 +00:00
mtx_assert(&sched_lock, MA_OWNED);
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
/* Normalize to zero. */
n = nice + SCHED_PRI_NHALF;
kseq->ksq_nice[n]--;
KASSERT(kseq->ksq_nice[n] >= 0, ("Negative nice count."));
/*
* If this wasn't the smallest nice value or there are more in
* this bucket we can just return. Otherwise we have to recalculate
* the smallest nice.
*/
if (nice != kseq->ksq_nicemin ||
kseq->ksq_nice[n] != 0 ||
- Remove the ksq_loads[] array. We are only interested in three counts, the total load, the timeshare load, and the number of threads that can be migrated to another cpu. Account for these seperately. - Introduce a KSE_CAN_MIGRATE() macro which determines whether or not a KSE can be migrated to another CPU. Currently, this only checks to see if we're an interrupt handler. Eventually this will also be used to support CPU binding.
2003-11-02 10:56:48 +00:00
kseq->ksq_load_timeshare == 0)
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
return;
- Remove uses of PRIO_TOTAL and replace them with SCHED_PRI_NRESV - SCHED_PRI_NRESV does not have the off by one error in PRIO_TOTAL so we do not have to account for it in the few places that we use it. Requested by: bde
2003-11-02 03:49:32 +00:00
for (; n < SCHED_PRI_NRESV; n++)
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
if (kseq->ksq_nice[n]) {
kseq->ksq_nicemin = n - SCHED_PRI_NHALF;
return;
}
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
}
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
#ifdef SMP
- Add a simple CPU load balancing algorithm. This works by executing once a second and equalizing the load between the two most imbalanced CPU. This is intended to clear up long term load imbalances that would not be handled by the 'pull' method in sched_choose(). - Pull out some bits of sched_choose() into a kseq_move() function that moves an arbitrary thread from one kseq to another.
2003-06-09 00:39:09 +00:00
/*
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
* sched_balance is a simple CPU load balancing algorithm. It operates by
- Add a simple CPU load balancing algorithm. This works by executing once a second and equalizing the load between the two most imbalanced CPU. This is intended to clear up long term load imbalances that would not be handled by the 'pull' method in sched_choose(). - Pull out some bits of sched_choose() into a kseq_move() function that moves an arbitrary thread from one kseq to another.
2003-06-09 00:39:09 +00:00
* finding the least loaded and most loaded cpu and equalizing their load
* by migrating some processes.
*
* Dealing only with two CPUs at a time has two advantages. Firstly, most
* installations will only have 2 cpus. Secondly, load balancing too much at
* once can have an unpleasant effect on the system. The scheduler rarely has
* enough information to make perfect decisions. So this algorithm chooses
* algorithm simplicity and more gradual effects on load in larger systems.
*
* It could be improved by considering the priorities and slices assigned to
* each task prior to balancing them. There are many pathological cases with
* any approach and so the semi random algorithm below may work as well as any.
*
*/
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
static void
- Run sched_balance() and sched_balance_groups() from hardclock via sched_clock() rather than using callouts. This means we no longer have to take the load of the callout thread into consideration while balancing and should make the balancing decisions simpler and more accurate. Tested on: x86/UP, amd64/SMP
2004-06-02 05:46:48 +00:00
sched_balance(void)
- Add a simple CPU load balancing algorithm. This works by executing once a second and equalizing the load between the two most imbalanced CPU. This is intended to clear up long term load imbalances that would not be handled by the 'pull' method in sched_choose(). - Pull out some bits of sched_choose() into a kseq_move() function that moves an arbitrary thread from one kseq to another.
2003-06-09 00:39:09 +00:00
{
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
struct kseq_group *high;
struct kseq_group *low;
struct kseq_group *ksg;
int cnt;
- Add a simple CPU load balancing algorithm. This works by executing once a second and equalizing the load between the two most imbalanced CPU. This is intended to clear up long term load imbalances that would not be handled by the 'pull' method in sched_choose(). - Pull out some bits of sched_choose() into a kseq_move() function that moves an arbitrary thread from one kseq to another.
2003-06-09 00:39:09 +00:00
int i;
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
bal_tick = ticks + (random() % (hz * 2));
- If smp is not started yet don't try to load balance or we'll put threads on cpus that aren't running yet.
2003-06-28 08:24:42 +00:00
if (smp_started == 0)
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
return;
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
low = high = NULL;
i = random() % (ksg_maxid + 1);
for (cnt = 0; cnt <= ksg_maxid; cnt++) {
ksg = KSEQ_GROUP(i);
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
/*
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
* Find the CPU with the highest load that has some
* threads to transfer.
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
*/
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
if ((high == NULL || ksg->ksg_load > high->ksg_load)
&& ksg->ksg_transferable)
high = ksg;
if (low == NULL || ksg->ksg_load < low->ksg_load)
low = ksg;
if (++i > ksg_maxid)
i = 0;
- Add a simple CPU load balancing algorithm. This works by executing once a second and equalizing the load between the two most imbalanced CPU. This is intended to clear up long term load imbalances that would not be handled by the 'pull' method in sched_choose(). - Pull out some bits of sched_choose() into a kseq_move() function that moves an arbitrary thread from one kseq to another.
2003-06-09 00:39:09 +00:00
}
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
if (low != NULL && high != NULL && high != low)
sched_balance_pair(LIST_FIRST(&high->ksg_members),
LIST_FIRST(&low->ksg_members));
}
static void
- Run sched_balance() and sched_balance_groups() from hardclock via sched_clock() rather than using callouts. This means we no longer have to take the load of the callout thread into consideration while balancing and should make the balancing decisions simpler and more accurate. Tested on: x86/UP, amd64/SMP
2004-06-02 05:46:48 +00:00
sched_balance_groups(void)
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
{
int i;
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
gbal_tick = ticks + (random() % (hz * 2));
- Run sched_balance() and sched_balance_groups() from hardclock via sched_clock() rather than using callouts. This means we no longer have to take the load of the callout thread into consideration while balancing and should make the balancing decisions simpler and more accurate. Tested on: x86/UP, amd64/SMP
2004-06-02 05:46:48 +00:00
mtx_assert(&sched_lock, MA_OWNED);
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
if (smp_started)
for (i = 0; i <= ksg_maxid; i++)
sched_balance_group(KSEQ_GROUP(i));
}
static void
sched_balance_group(struct kseq_group *ksg)
{
struct kseq *kseq;
struct kseq *high;
struct kseq *low;
int load;
if (ksg->ksg_transferable == 0)
return;
low = NULL;
high = NULL;
LIST_FOREACH(kseq, &ksg->ksg_members, ksq_siblings) {
load = kseq->ksq_load;
if (high == NULL || load > high->ksq_load)
high = kseq;
if (low == NULL || load < low->ksq_load)
low = kseq;
}
if (high != NULL && low != NULL && high != low)
sched_balance_pair(high, low);
}
static void
sched_balance_pair(struct kseq *high, struct kseq *low)
{
int transferable;
int high_load;
int low_load;
int move;
int diff;
int i;
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
/*
* If we're transfering within a group we have to use this specific
* kseq's transferable count, otherwise we can steal from other members
* of the group.
*/
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
if (high->ksq_group == low->ksq_group) {
transferable = high->ksq_transferable;
high_load = high->ksq_load;
low_load = low->ksq_load;
} else {
transferable = high->ksq_group->ksg_transferable;
high_load = high->ksq_group->ksg_load;
low_load = low->ksq_group->ksg_load;
}
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
if (transferable == 0)
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
return;
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
/*
* Determine what the imbalance is and then adjust that to how many
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
* kses we actually have to give up (transferable).
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
*/
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
diff = high_load - low_load;
- Add a simple CPU load balancing algorithm. This works by executing once a second and equalizing the load between the two most imbalanced CPU. This is intended to clear up long term load imbalances that would not be handled by the 'pull' method in sched_choose(). - Pull out some bits of sched_choose() into a kseq_move() function that moves an arbitrary thread from one kseq to another.
2003-06-09 00:39:09 +00:00
move = diff / 2;
if (diff & 0x1)
move++;
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
move = min(move, transferable);
- Add a simple CPU load balancing algorithm. This works by executing once a second and equalizing the load between the two most imbalanced CPU. This is intended to clear up long term load imbalances that would not be handled by the 'pull' method in sched_choose(). - Pull out some bits of sched_choose() into a kseq_move() function that moves an arbitrary thread from one kseq to another.
2003-06-09 00:39:09 +00:00
for (i = 0; i < move; i++)
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
kseq_move(high, KSEQ_ID(low));
- Add a simple CPU load balancing algorithm. This works by executing once a second and equalizing the load between the two most imbalanced CPU. This is intended to clear up long term load imbalances that would not be handled by the 'pull' method in sched_choose(). - Pull out some bits of sched_choose() into a kseq_move() function that moves an arbitrary thread from one kseq to another.
2003-06-09 00:39:09 +00:00
return;
}
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
static void
- Add a simple CPU load balancing algorithm. This works by executing once a second and equalizing the load between the two most imbalanced CPU. This is intended to clear up long term load imbalances that would not be handled by the 'pull' method in sched_choose(). - Pull out some bits of sched_choose() into a kseq_move() function that moves an arbitrary thread from one kseq to another.
2003-06-09 00:39:09 +00:00
kseq_move(struct kseq *from, int cpu)
{
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
struct kseq *kseq;
struct kseq *to;
- Add a simple CPU load balancing algorithm. This works by executing once a second and equalizing the load between the two most imbalanced CPU. This is intended to clear up long term load imbalances that would not be handled by the 'pull' method in sched_choose(). - Pull out some bits of sched_choose() into a kseq_move() function that moves an arbitrary thread from one kseq to another.
2003-06-09 00:39:09 +00:00
struct kse *ke;
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
kseq = from;
to = KSEQ_CPU(cpu);
ke = kseq_steal(kseq, 1);
if (ke == NULL) {
struct kseq_group *ksg;
ksg = kseq->ksq_group;
LIST_FOREACH(kseq, &ksg->ksg_members, ksq_siblings) {
if (kseq == from || kseq->ksq_transferable == 0)
continue;
ke = kseq_steal(kseq, 1);
break;
}
if (ke == NULL)
panic("kseq_move: No KSEs available with a "
"transferable count of %d\n",
ksg->ksg_transferable);
}
if (kseq == to)
return;
- Add a simple CPU load balancing algorithm. This works by executing once a second and equalizing the load between the two most imbalanced CPU. This is intended to clear up long term load imbalances that would not be handled by the 'pull' method in sched_choose(). - Pull out some bits of sched_choose() into a kseq_move() function that moves an arbitrary thread from one kseq to another.
2003-06-09 00:39:09 +00:00
ke->ke_state = KES_THREAD;
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
kseq_runq_rem(kseq, ke);
kseq_load_rem(kseq, ke);
- Remove kseq_find(), we no longer scan other cpu's run queues when we go idle. They figure out that we're idle fast enough that the cache pollution introduces by scanning their run queue is more expensive than waiting a little longer. - Add kseq_setidle() to mark us as being idle. Use this in place of kseq_find(). - Remove kseq_load_highest(), kseq_find() was the only consumer of this interface. kseq_balance() has it's own customized version that finds the lowest and highest loads simultaneously. Continuously told that this would be faster by: terry
2003-11-03 03:27:22 +00:00
kseq_notify(ke, cpu);
- Add a simple CPU load balancing algorithm. This works by executing once a second and equalizing the load between the two most imbalanced CPU. This is intended to clear up long term load imbalances that would not be handled by the 'pull' method in sched_choose(). - Pull out some bits of sched_choose() into a kseq_move() function that moves an arbitrary thread from one kseq to another.
2003-06-09 00:39:09 +00:00
}
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
static int
kseq_idled(struct kseq *kseq)
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
{
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
struct kseq_group *ksg;
struct kseq *steal;
struct kse *ke;
ksg = kseq->ksq_group;
/*
* If we're in a cpu group, try and steal kses from another cpu in
* the group before idling.
*/
if (ksg->ksg_cpus > 1 && ksg->ksg_transferable) {
LIST_FOREACH(steal, &ksg->ksg_members, ksq_siblings) {
if (steal == kseq || steal->ksq_transferable == 0)
continue;
ke = kseq_steal(steal, 0);
if (ke == NULL)
continue;
ke->ke_state = KES_THREAD;
kseq_runq_rem(steal, ke);
kseq_load_rem(steal, ke);
ke->ke_cpu = PCPU_GET(cpuid);
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
ke->ke_flags |= KEF_INTERNAL | KEF_HOLD;
sched_add(ke->ke_thread, SRQ_YIELDING);
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
return (0);
}
}
/*
* We only set the idled bit when all of the cpus in the group are
* idle. Otherwise we could get into a situation where a KSE bounces
* back and forth between two idle cores on seperate physical CPUs.
*/
ksg->ksg_idlemask |= PCPU_GET(cpumask);
if (ksg->ksg_idlemask != ksg->ksg_cpumask)
return (1);
atomic_set_int(&kseq_idle, ksg->ksg_mask);
return (1);
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
}
static void
kseq_assign(struct kseq *kseq)
{
struct kse *nke;
struct kse *ke;
do {
Avoid casts as lvalues.
2004-07-28 06:42:41 +00:00
*(volatile struct kse **)&ke = kseq->ksq_assigned;
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
} while(!atomic_cmpset_ptr(&kseq->ksq_assigned, ke, NULL));
for (; ke != NULL; ke = nke) {
nke = ke->ke_assign;
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
kseq->ksq_group->ksg_load--;
kseq->ksq_load--;
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
ke->ke_flags &= ~KEF_ASSIGNED;
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
ke->ke_flags |= KEF_INTERNAL | KEF_HOLD;
sched_add(ke->ke_thread, SRQ_YIELDING);
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
}
}
static void
kseq_notify(struct kse *ke, int cpu)
{
struct kseq *kseq;
struct thread *td;
struct pcpu *pcpu;
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
int class;
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
int prio;
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
kseq = KSEQ_CPU(cpu);
/* XXX */
class = PRI_BASE(ke->ke_ksegrp->kg_pri_class);
if ((class == PRI_TIMESHARE || class == PRI_REALTIME) &&
(kseq_idle & kseq->ksq_group->ksg_mask))
atomic_clear_int(&kseq_idle, kseq->ksq_group->ksg_mask);
kseq->ksq_group->ksg_load++;
kseq->ksq_load++;
- Assign the ke_cpu field in kseq_notify() so that all of our callers do not have to do it. - Set the ke_runq to NULL in sched_add() before calling kseq_notify(). Otherwise we may panic in sched_add() if INVARIANTS is on.
2003-12-14 02:06:29 +00:00
ke->ke_cpu = cpu;
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
ke->ke_flags |= KEF_ASSIGNED;
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
prio = ke->ke_thread->td_priority;
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
/*
* Place a KSE on another cpu's queue and force a resched.
*/
do {
Avoid casts as lvalues.
2004-07-28 06:42:41 +00:00
*(volatile struct kse **)&ke->ke_assign = kseq->ksq_assigned;
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
} while(!atomic_cmpset_ptr(&kseq->ksq_assigned, ke->ke_assign, ke));
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
/*
* Without sched_lock we could lose a race where we set NEEDRESCHED
* on a thread that is switched out before the IPI is delivered. This
* would lead us to miss the resched. This will be a problem once
* sched_lock is pushed down.
*/
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
pcpu = pcpu_find(cpu);
td = pcpu->pc_curthread;
if (ke->ke_thread->td_priority < td->td_priority ||
td == pcpu->pc_idlethread) {
td->td_flags |= TDF_NEEDRESCHED;
ipi_selected(1 << cpu, IPI_AST);
}
}
static struct kse *
runq_steal(struct runq *rq)
{
struct rqhead *rqh;
struct rqbits *rqb;
struct kse *ke;
int word;
int bit;
mtx_assert(&sched_lock, MA_OWNED);
rqb = &rq->rq_status;
for (word = 0; word < RQB_LEN; word++) {
if (rqb->rqb_bits[word] == 0)
continue;
for (bit = 0; bit < RQB_BPW; bit++) {
rqb_bits[] may be an int64_t (eg: on alpha, and recently on amd64). Be sure to shift (long)1 << 33 and higher, not (int)1. Otherwise bad things happen(TM). This is why beast.freebsd.org paniced with ULE. Reviewed by: jeff
2003-12-07 09:57:51 +00:00
if ((rqb->rqb_bits[word] & (1ul << bit)) == 0)
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
continue;
rqh = &rq->rq_queues[bit + (word << RQB_L2BPW)];
TAILQ_FOREACH(ke, rqh, ke_procq) {
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
if (KSE_CAN_MIGRATE(ke))
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
return (ke);
}
}
}
return (NULL);
}
static struct kse *
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
kseq_steal(struct kseq *kseq, int stealidle)
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
{
struct kse *ke;
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
/*
* Steal from next first to try to get a non-interactive task that
* may not have run for a while.
*/
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
if ((ke = runq_steal(kseq->ksq_next)) != NULL)
return (ke);
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
if ((ke = runq_steal(kseq->ksq_curr)) != NULL)
return (ke);
if (stealidle)
return (runq_steal(&kseq->ksq_idle));
return (NULL);
}
int
kseq_transfer(struct kseq *kseq, struct kse *ke, int class)
{
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
struct kseq_group *nksg;
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
struct kseq_group *ksg;
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
struct kseq *old;
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
int cpu;
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
int idx;
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
- In kseq_transfer() return if smp has not been started. - In sched_add(), do the idle check prior to the transfer check so that we don't try to transfer load from an idle cpu. This fixes panics caused by IPIs on UP machines running SMP kernels. Reported/Debugged by: seanc
2003-12-20 14:03:14 +00:00
if (smp_started == 0)
return (0);
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
cpu = 0;
- In kseq_transfer() return if smp has not been started. - In sched_add(), do the idle check prior to the transfer check so that we don't try to transfer load from an idle cpu. This fixes panics caused by IPIs on UP machines running SMP kernels. Reported/Debugged by: seanc
2003-12-20 14:03:14 +00:00
/*
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
* If our load exceeds a certain threshold we should attempt to
* reassign this thread. The first candidate is the cpu that
* originally ran the thread. If it is idle, assign it there,
* otherwise, pick an idle cpu.
*
* The threshold at which we start to reassign kses has a large impact
- In kseq_transfer() return if smp has not been started. - In sched_add(), do the idle check prior to the transfer check so that we don't try to transfer load from an idle cpu. This fixes panics caused by IPIs on UP machines running SMP kernels. Reported/Debugged by: seanc
2003-12-20 14:03:14 +00:00
* on the overall performance of the system. Tuned too high and
* some CPUs may idle. Too low and there will be excess migration
Spell "switches" a more conventional way.
2004-04-09 14:31:29 +00:00
* and context switches.
- In kseq_transfer() return if smp has not been started. - In sched_add(), do the idle check prior to the transfer check so that we don't try to transfer load from an idle cpu. This fixes panics caused by IPIs on UP machines running SMP kernels. Reported/Debugged by: seanc
2003-12-20 14:03:14 +00:00
*/
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
old = KSEQ_CPU(ke->ke_cpu);
nksg = old->ksq_group;
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
ksg = kseq->ksq_group;
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
if (kseq_idle) {
if (kseq_idle & nksg->ksg_mask) {
cpu = ffs(nksg->ksg_idlemask);
if (cpu) {
CTR2(KTR_SCHED,
"kseq_transfer: %p found old cpu %X "
"in idlemask.", ke, cpu);
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
goto migrate;
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
}
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
}
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
/*
* Multiple cpus could find this bit simultaneously
* but the race shouldn't be terrible.
*/
cpu = ffs(kseq_idle);
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
if (cpu) {
CTR2(KTR_SCHED, "kseq_transfer: %p found %X "
"in idlemask.", ke, cpu);
goto migrate;
}
}
idx = 0;
#if 0
if (old->ksq_load < kseq->ksq_load) {
cpu = ke->ke_cpu + 1;
CTR2(KTR_SCHED, "kseq_transfer: %p old cpu %X "
"load less than ours.", ke, cpu);
goto migrate;
}
/*
* No new CPU was found, look for one with less load.
*/
for (idx = 0; idx <= ksg_maxid; idx++) {
nksg = KSEQ_GROUP(idx);
if (nksg->ksg_load /*+ (nksg->ksg_cpus * 2)*/ < ksg->ksg_load) {
cpu = ffs(nksg->ksg_cpumask);
CTR2(KTR_SCHED, "kseq_transfer: %p cpu %X load less "
"than ours.", ke, cpu);
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
goto migrate;
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
}
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
}
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
#endif
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
/*
* If another cpu in this group has idled, assign a thread over
* to them after checking to see if there are idled groups.
*/
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
if (ksg->ksg_idlemask) {
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
cpu = ffs(ksg->ksg_idlemask);
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
if (cpu) {
CTR2(KTR_SCHED, "kseq_transfer: %p cpu %X idle in "
"group.", ke, cpu);
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
goto migrate;
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
}
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
}
return (0);
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
migrate:
/*
* Now that we've found an idle CPU, migrate the thread.
*/
cpu--;
ke->ke_runq = NULL;
kseq_notify(ke, cpu);
return (1);
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
}
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
#endif /* SMP */
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
- When stealing a kse in kseq_move() ignore the current kseq's min nice value. We want to steal any thread, even one that is not given a slice on its current queue.
2003-07-08 06:19:40 +00:00
/*
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
* Pick the highest priority task we have and return it.
- When stealing a kse in kseq_move() ignore the current kseq's min nice value. We want to steal any thread, even one that is not given a slice on its current queue.
2003-07-08 06:19:40 +00:00
*/
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
static struct kse *
kseq_choose(struct kseq *kseq)
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
{
struct runq *swap;
- When choosing a thread on the run queue, check to see if its nice is outside of the nice threshold due to a recently awoken thread with a lower nice value. This further reduces the amount of time a positively niced thread gets while running in conjunction with a workload that has many short sleeps (ie buildworld).
2004-10-30 12:19:15 +00:00
struct kse *ke;
int nice;
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
- When a new thread is added to a kseq the load is incremented prior to adding it to the nice tables. Therefore, in kseq_add_nice, we should keep in mind that the load will be 1 if we are the only thread, and not 0. - Assert that the sched lock is held in all the appropriate places. - Increase the scope of the sched lock in sched_pctcpu_update(). - Hold the sched lock in sched_runnable(). It is not held by the caller.
2003-06-08 00:47:33 +00:00
mtx_assert(&sched_lock, MA_OWNED);
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
swap = NULL;
for (;;) {
ke = runq_choose(kseq->ksq_curr);
if (ke == NULL) {
/*
- Change mi_switch() and sched_switch() to accept an optional thread to switch to. If a non-NULL thread pointer is passed in, then the CPU will switch to that thread directly rather than calling choosethread() to pick a thread to choose to. - Make sched_switch() aware of idle threads and know to do TD_SET_CAN_RUN() instead of sticking them on the run queue rather than requiring all callers of mi_switch() to know to do this if they can be called from an idlethread. - Move constants for arguments to mi_switch() and thread_single() out of the middle of the function prototypes and up above into their own section.
2004-07-02 19:09:50 +00:00
* We already swapped once and didn't get anywhere.
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
*/
if (swap)
break;
swap = kseq->ksq_curr;
kseq->ksq_curr = kseq->ksq_next;
kseq->ksq_next = swap;
continue;
}
/*
* If we encounter a slice of 0 the kse is in a
* TIMESHARE kse group and its nice was too far out
* of the range that receives slices.
*/
- When choosing a thread on the run queue, check to see if its nice is outside of the nice threshold due to a recently awoken thread with a lower nice value. This further reduces the amount of time a positively niced thread gets while running in conjunction with a workload that has many short sleeps (ie buildworld).
2004-10-30 12:19:15 +00:00
nice = ke->ke_proc->p_nice + (0 - kseq->ksq_nicemin);
- In kseq_choose(), don't recalculate slice values for processes with a nice of 0. Doing so can cause an infinite loop because they should be running, but a nice -20 process could prevent them from doing so. - Add a new flag KEF_PRIOELEV to flag a thread that has had its priority elevated due to priority propagation. If a thread has had its priority elevated, we assume that it must go on the current queue and it must get a slice. - In sched_userret() if our priority was elevated and we shouldn't have a timeslice, yield here until we should. Found/Tested by: glebius
2004-12-14 10:34:27 +00:00
if (ke->ke_slice == 0 || (nice > SCHED_SLICE_NTHRESH &&
ke->ke_proc->p_nice != 0)) {
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
runq_remove(ke->ke_runq, ke);
sched_slice(ke);
ke->ke_runq = kseq->ksq_next;
When preempting a thread, put it back on the HEAD of its run queue. (Only really implemented in 4bsd) MFC after: 4 days
2004-10-05 22:03:10 +00:00
runq_add(ke->ke_runq, ke, 0);
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
continue;
}
return (ke);
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
}
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
return (runq_choose(&kseq->ksq_idle));
- Add in support for KSEs with 0 slice values on the run queue. If we try to select a KSE with a slice of 0 we will update its slice and insert it onto the next queue. - Pass the KSE instead of the ksegrp into sched_slice(). This more accurately reflects the behavior of the code. Slices are granted to kses. - Add a function kseq_nice_min() which finds the smallest nice value assigned to the kseg of any KSE on the queue. - Rewrite the logic in sched_slice(). Add a large comment describing the new slice selection scheme. To summarize, slices are assigned based on the nice value. Priorities are still calculated based on the nice and interactivity of a process. Slice sizes of 0 may be granted for KSEs whos nice is 20 or futher away from the lowest nice on the run queue. Other nice values are scaled across the range [min, min+20]. This fixes ULEs bad behavior with positively niced processes.
2003-04-02 06:46:43 +00:00
}
- Use ksq_load as the authoritive count of kses on the pair of kseqs for sched_runnable() et all. - Remove some dead code in sched_clock(). - Define two macros KSEQ_SELF() and KSEQ_CPU() for getting the kseq of the current cpu or some alternate cpu. - Start introducing kseq_() functions, such as kseq_choose() and kseq_setup().
2003-01-29 07:00:51 +00:00
static void
kseq_setup(struct kseq *kseq)
{
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
runq_init(&kseq->ksq_timeshare[0]);
runq_init(&kseq->ksq_timeshare[1]);
- Keep seperate statistics and run queues for different scheduling classes. - Treat each class specially in kseq_{choose,add,rem}. Let the rest of the code be less aware of scheduling classes. - Skip the interactivity calculation for non TIMESHARE ksegrps. - Move slice and runq selection into kseq_add(). Uninline it now that it's big.
2003-04-03 00:29:28 +00:00
runq_init(&kseq->ksq_idle);
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
kseq->ksq_curr = &kseq->ksq_timeshare[0];
kseq->ksq_next = &kseq->ksq_timeshare[1];
- Set the ke_cpu field in sched_add() for interrupt and realtime threads since they are going on the current cpu and not their previously assigned cpu. - sched_runnable() should only return true in the SMP case if the other processor has more than one thread that is runnable. We can not steal curthread. - Change kseq_print() to accept the cpuid instead of a kseq pointer. This makes use of this function in ddb much easier.
2003-04-18 05:24:10 +00:00
kseq->ksq_load = 0;
- Remove the ksq_loads[] array. We are only interested in three counts, the total load, the timeshare load, and the number of threads that can be migrated to another cpu. Account for these seperately. - Introduce a KSE_CAN_MIGRATE() macro which determines whether or not a KSE can be migrated to another CPU. Currently, this only checks to see if we're an interrupt handler. Eventually this will also be used to support CPU binding.
2003-11-02 10:56:48 +00:00
kseq->ksq_load_timeshare = 0;
- Use ksq_load as the authoritive count of kses on the pair of kseqs for sched_runnable() et all. - Remove some dead code in sched_clock(). - Define two macros KSEQ_SELF() and KSEQ_CPU() for getting the kseq of the current cpu or some alternate cpu. - Start introducing kseq_() functions, such as kseq_choose() and kseq_setup().
2003-01-29 07:00:51 +00:00
}
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
static void
sched_setup(void *dummy)
{
- Clean up an unused variable. Submitted by: Steve Kargl <skg@routmask.apl.washington.edu>
2003-07-07 21:08:28 +00:00
#ifdef SMP
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
int i;
- Clean up an unused variable. Submitted by: Steve Kargl <skg@routmask.apl.washington.edu>
2003-07-07 21:08:28 +00:00
#endif
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
- Adjust the default maximum slice value to ~140ms. This has improved the nice distribution without significantly impacting interactive response. As a side effect it should also allow batch processes to run for a slightly longer period which will positively impact their performance.
2003-06-28 06:04:47 +00:00
slice_min = (hz/100); /* 10ms */
slice_max = (hz/7); /* ~140ms */
- Create a function sched_interact_score() which decides on the interactivity of a kseg and assigns it a value of 0 through 100. - Use sched_interact_score() to determine the dynamic priority. - Define SCHED_CURR() in terms of sched_interact_score(). - Adjust the maximum slice back down to 100ms. - Remove redundant clearing of ke_runq in sched_wakeup() - Clean up #defines and comment them.
2003-03-04 02:45:59 +00:00
- Parse the cpu topology map in sched_setup(). - Associate logical CPUs on the same physical core with the same kseq. - Adjust code that assumed there would only be one running thread in any kseq. - Wrap the HTT code with a ULE_HTT_EXPERIMENTAL ifdef. This is a start towards HyperThreading support but it isn't quite there yet.
2003-07-04 19:59:00 +00:00
#ifdef SMP
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
balance_groups = 0;
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
/*
* Initialize the kseqs.
*/
for (i = 0; i < MAXCPU; i++) {
struct kseq *ksq;
ksq = &kseq_cpu[i];
ksq->ksq_assigned = NULL;
kseq_setup(&kseq_cpu[i]);
}
- Parse the cpu topology map in sched_setup(). - Associate logical CPUs on the same physical core with the same kseq. - Adjust code that assumed there would only be one running thread in any kseq. - Wrap the HTT code with a ULE_HTT_EXPERIMENTAL ifdef. This is a start towards HyperThreading support but it isn't quite there yet.
2003-07-04 19:59:00 +00:00
if (smp_topology == NULL) {
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
struct kseq_group *ksg;
struct kseq *ksq;
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
int cpus;
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
for (cpus = 0, i = 0; i < MAXCPU; i++) {
if (CPU_ABSENT(i))
continue;
ksq = &kseq_cpu[cpus];
ksg = &kseq_groups[cpus];
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
/*
- Run sched_balance() and sched_balance_groups() from hardclock via sched_clock() rather than using callouts. This means we no longer have to take the load of the callout thread into consideration while balancing and should make the balancing decisions simpler and more accurate. Tested on: x86/UP, amd64/SMP
2004-06-02 05:46:48 +00:00
* Setup a kseq group with one member.
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
*/
ksq->ksq_transferable = 0;
ksq->ksq_group = ksg;
ksg->ksg_cpus = 1;
ksg->ksg_idlemask = 0;
ksg->ksg_cpumask = ksg->ksg_mask = 1 << i;
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
ksg->ksg_load = 0;
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
ksg->ksg_transferable = 0;
LIST_INIT(&ksg->ksg_members);
LIST_INSERT_HEAD(&ksg->ksg_members, ksq, ksq_siblings);
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
cpus++;
- Parse the cpu topology map in sched_setup(). - Associate logical CPUs on the same physical core with the same kseq. - Adjust code that assumed there would only be one running thread in any kseq. - Wrap the HTT code with a ULE_HTT_EXPERIMENTAL ifdef. This is a start towards HyperThreading support but it isn't quite there yet.
2003-07-04 19:59:00 +00:00
}
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
ksg_maxid = cpus - 1;
- Parse the cpu topology map in sched_setup(). - Associate logical CPUs on the same physical core with the same kseq. - Adjust code that assumed there would only be one running thread in any kseq. - Wrap the HTT code with a ULE_HTT_EXPERIMENTAL ifdef. This is a start towards HyperThreading support but it isn't quite there yet.
2003-07-04 19:59:00 +00:00
} else {
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
struct kseq_group *ksg;
struct cpu_group *cg;
- Parse the cpu topology map in sched_setup(). - Associate logical CPUs on the same physical core with the same kseq. - Adjust code that assumed there would only be one running thread in any kseq. - Wrap the HTT code with a ULE_HTT_EXPERIMENTAL ifdef. This is a start towards HyperThreading support but it isn't quite there yet.
2003-07-04 19:59:00 +00:00
int j;
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
- Parse the cpu topology map in sched_setup(). - Associate logical CPUs on the same physical core with the same kseq. - Adjust code that assumed there would only be one running thread in any kseq. - Wrap the HTT code with a ULE_HTT_EXPERIMENTAL ifdef. This is a start towards HyperThreading support but it isn't quite there yet.
2003-07-04 19:59:00 +00:00
for (i = 0; i < smp_topology->ct_count; i++) {
cg = &smp_topology->ct_group[i];
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
ksg = &kseq_groups[i];
/*
* Initialize the group.
*/
ksg->ksg_idlemask = 0;
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
ksg->ksg_load = 0;
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
ksg->ksg_transferable = 0;
ksg->ksg_cpus = cg->cg_count;
ksg->ksg_cpumask = cg->cg_mask;
LIST_INIT(&ksg->ksg_members);
/*
* Find all of the group members and add them.
*/
for (j = 0; j < MAXCPU; j++) {
if ((cg->cg_mask & (1 << j)) != 0) {
if (ksg->ksg_mask == 0)
ksg->ksg_mask = 1 << j;
kseq_cpu[j].ksq_transferable = 0;
kseq_cpu[j].ksq_group = ksg;
LIST_INSERT_HEAD(&ksg->ksg_members,
&kseq_cpu[j], ksq_siblings);
}
}
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
if (ksg->ksg_cpus > 1)
balance_groups = 1;
- Parse the cpu topology map in sched_setup(). - Associate logical CPUs on the same physical core with the same kseq. - Adjust code that assumed there would only be one running thread in any kseq. - Wrap the HTT code with a ULE_HTT_EXPERIMENTAL ifdef. This is a start towards HyperThreading support but it isn't quite there yet.
2003-07-04 19:59:00 +00:00
}
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
ksg_maxid = smp_topology->ct_count - 1;
- Parse the cpu topology map in sched_setup(). - Associate logical CPUs on the same physical core with the same kseq. - Adjust code that assumed there would only be one running thread in any kseq. - Wrap the HTT code with a ULE_HTT_EXPERIMENTAL ifdef. This is a start towards HyperThreading support but it isn't quite there yet.
2003-07-04 19:59:00 +00:00
}
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
/*
* Stagger the group and global load balancer so they do not
* interfere with each other.
*/
- Run sched_balance() and sched_balance_groups() from hardclock via sched_clock() rather than using callouts. This means we no longer have to take the load of the callout thread into consideration while balancing and should make the balancing decisions simpler and more accurate. Tested on: x86/UP, amd64/SMP
2004-06-02 05:46:48 +00:00
bal_tick = ticks + hz;
- Now that we have kseq groups, balance them seperately. - The new sched_balance_groups() function does intra-group balancing while sched_balance() balances the available groups. - Pick a random time between 0 ticks and hz * 2 ticks to restart each balancing process. Each balancer has its own timeout. - Pick a random place in the list of groups to start the search for lowest and highest group loads. This prevents us from prefering a group based on numeric position. - Use a nasty hack to stop us from preferring cpu 0. The problem is that softclock always runs on cpu 0, so it always has a little extra load. We ignore this load in the balancer for now. In the future softclock should run on a random cpu and these hacks can go away.
2003-12-12 07:33:51 +00:00
if (balance_groups)
- Run sched_balance() and sched_balance_groups() from hardclock via sched_clock() rather than using callouts. This means we no longer have to take the load of the callout thread into consideration while balancing and should make the balancing decisions simpler and more accurate. Tested on: x86/UP, amd64/SMP
2004-06-02 05:46:48 +00:00
gbal_tick = ticks + (hz / 2);
- Parse the cpu topology map in sched_setup(). - Associate logical CPUs on the same physical core with the same kseq. - Adjust code that assumed there would only be one running thread in any kseq. - Wrap the HTT code with a ULE_HTT_EXPERIMENTAL ifdef. This is a start towards HyperThreading support but it isn't quite there yet.
2003-07-04 19:59:00 +00:00
#else
kseq_setup(KSEQ_SELF());
- Add a simple CPU load balancing algorithm. This works by executing once a second and equalizing the load between the two most imbalanced CPU. This is intended to clear up long term load imbalances that would not be handled by the 'pull' method in sched_choose(). - Pull out some bits of sched_choose() into a kseq_move() function that moves an arbitrary thread from one kseq to another.
2003-06-09 00:39:09 +00:00
#endif
- Parse the cpu topology map in sched_setup(). - Associate logical CPUs on the same physical core with the same kseq. - Adjust code that assumed there would only be one running thread in any kseq. - Wrap the HTT code with a ULE_HTT_EXPERIMENTAL ifdef. This is a start towards HyperThreading support but it isn't quite there yet.
2003-07-04 19:59:00 +00:00
mtx_lock_spin(&sched_lock);
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
kseq_load_add(KSEQ_SELF(), &kse0);
- Parse the cpu topology map in sched_setup(). - Associate logical CPUs on the same physical core with the same kseq. - Adjust code that assumed there would only be one running thread in any kseq. - Wrap the HTT code with a ULE_HTT_EXPERIMENTAL ifdef. This is a start towards HyperThreading support but it isn't quite there yet.
2003-07-04 19:59:00 +00:00
mtx_unlock_spin(&sched_lock);
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
}
/*
* Scale the scheduling priority according to the "interactivity" of this
* process.
*/
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
static void
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
sched_priority(struct ksegrp *kg)
{
int pri;
if (kg->kg_pri_class != PRI_TIMESHARE)
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
return;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
pri = SCHED_PRI_INTERACT(sched_interact_score(kg));
- Create a function sched_interact_score() which decides on the interactivity of a kseg and assigns it a value of 0 through 100. - Use sched_interact_score() to determine the dynamic priority. - Define SCHED_CURR() in terms of sched_interact_score(). - Adjust the maximum slice back down to 100ms. - Remove redundant clearing of ke_runq in sched_wakeup() - Clean up #defines and comment them.
2003-03-04 02:45:59 +00:00
pri += SCHED_PRI_BASE;
Nice, is a property of a process as a whole.. I mistakenly moved it to the ksegroup when breaking up the process structure. Put it back in the proc structure.
2004-06-16 00:26:31 +00:00
pri += kg->kg_proc->p_nice;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
if (pri > PRI_MAX_TIMESHARE)
pri = PRI_MAX_TIMESHARE;
else if (pri < PRI_MIN_TIMESHARE)
pri = PRI_MIN_TIMESHARE;
kg->kg_user_pri = pri;
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
return;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
}
/*
- Add in support for KSEs with 0 slice values on the run queue. If we try to select a KSE with a slice of 0 we will update its slice and insert it onto the next queue. - Pass the KSE instead of the ksegrp into sched_slice(). This more accurately reflects the behavior of the code. Slices are granted to kses. - Add a function kseq_nice_min() which finds the smallest nice value assigned to the kseg of any KSE on the queue. - Rewrite the logic in sched_slice(). Add a large comment describing the new slice selection scheme. To summarize, slices are assigned based on the nice value. Priorities are still calculated based on the nice and interactivity of a process. Slice sizes of 0 may be granted for KSEs whos nice is 20 or futher away from the lowest nice on the run queue. Other nice values are scaled across the range [min, min+20]. This fixes ULEs bad behavior with positively niced processes.
2003-04-02 06:46:43 +00:00
* Calculate a time slice based on the properties of the kseg and the runq
- Keep seperate statistics and run queues for different scheduling classes. - Treat each class specially in kseq_{choose,add,rem}. Let the rest of the code be less aware of scheduling classes. - Skip the interactivity calculation for non TIMESHARE ksegrps. - Move slice and runq selection into kseq_add(). Uninline it now that it's big.
2003-04-03 00:29:28 +00:00
* that we're on. This is only for PRI_TIMESHARE ksegrps.
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
*/
- Add in support for KSEs with 0 slice values on the run queue. If we try to select a KSE with a slice of 0 we will update its slice and insert it onto the next queue. - Pass the KSE instead of the ksegrp into sched_slice(). This more accurately reflects the behavior of the code. Slices are granted to kses. - Add a function kseq_nice_min() which finds the smallest nice value assigned to the kseg of any KSE on the queue. - Rewrite the logic in sched_slice(). Add a large comment describing the new slice selection scheme. To summarize, slices are assigned based on the nice value. Priorities are still calculated based on the nice and interactivity of a process. Slice sizes of 0 may be granted for KSEs whos nice is 20 or futher away from the lowest nice on the run queue. Other nice values are scaled across the range [min, min+20]. This fixes ULEs bad behavior with positively niced processes.
2003-04-02 06:46:43 +00:00
static void
sched_slice(struct kse *ke)
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
{
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
struct kseq *kseq;
- Add in support for KSEs with 0 slice values on the run queue. If we try to select a KSE with a slice of 0 we will update its slice and insert it onto the next queue. - Pass the KSE instead of the ksegrp into sched_slice(). This more accurately reflects the behavior of the code. Slices are granted to kses. - Add a function kseq_nice_min() which finds the smallest nice value assigned to the kseg of any KSE on the queue. - Rewrite the logic in sched_slice(). Add a large comment describing the new slice selection scheme. To summarize, slices are assigned based on the nice value. Priorities are still calculated based on the nice and interactivity of a process. Slice sizes of 0 may be granted for KSEs whos nice is 20 or futher away from the lowest nice on the run queue. Other nice values are scaled across the range [min, min+20]. This fixes ULEs bad behavior with positively niced processes.
2003-04-02 06:46:43 +00:00
struct ksegrp *kg;
kg = ke->ke_ksegrp;
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
kseq = KSEQ_CPU(ke->ke_cpu);
- Add in support for KSEs with 0 slice values on the run queue. If we try to select a KSE with a slice of 0 we will update its slice and insert it onto the next queue. - Pass the KSE instead of the ksegrp into sched_slice(). This more accurately reflects the behavior of the code. Slices are granted to kses. - Add a function kseq_nice_min() which finds the smallest nice value assigned to the kseg of any KSE on the queue. - Rewrite the logic in sched_slice(). Add a large comment describing the new slice selection scheme. To summarize, slices are assigned based on the nice value. Priorities are still calculated based on the nice and interactivity of a process. Slice sizes of 0 may be granted for KSEs whos nice is 20 or futher away from the lowest nice on the run queue. Other nice values are scaled across the range [min, min+20]. This fixes ULEs bad behavior with positively niced processes.
2003-04-02 06:46:43 +00:00
Rework the interface between priority propagation (lending) and the schedulers a bit to ensure more correct handling of priorities and fewer priority inversions: - Add two functions to the sched(9) API to handle priority lending: sched_lend_prio() and sched_unlend_prio(). The turnstile code uses these functions to ask the scheduler to lend a thread a set priority and to tell the scheduler when it thinks it is ok for a thread to stop borrowing priority. The unlend case is slightly complex in that the turnstile code tells the scheduler what the minimum priority of the thread needs to be to satisfy the requirements of any other threads blocked on locks owned by the thread in question. The scheduler then decides where the thread can go back to normal mode (if it's normal priority is high enough to satisfy the pending lock requests) or it it should continue to use the priority specified to the sched_unlend_prio() call. This involves adding a new per-thread flag TDF_BORROWING that replaces the ULE-only kse flag for priority elevation. - Schedulers now refuse to lower the priority of a thread that is currently borrowing another therad's priority. - If a scheduler changes the priority of a thread that is currently sitting on a turnstile, it will call a new function turnstile_adjust() to inform the turnstile code of the change. This function resorts the thread on the priority list of the turnstile if needed, and if the thread ends up at the head of the list (due to having the highest priority) and its priority was raised, then it will propagate that new priority to the owner of the lock it is blocked on. Some additional fixes specific to the 4BSD scheduler include: - Common code for updating the priority of a thread when the user priority of its associated kse group has been consolidated in a new static function resetpriority_thread(). One change to this function is that it will now only adjust the priority of a thread if it already has a time sharing priority, thus preserving any boosts from a tsleep() until the thread returns to userland. Also, resetpriority() no longer calls maybe_resched() on each thread in the group. Instead, the code calling resetpriority() is responsible for calling resetpriority_thread() on any threads that need to be updated. - schedcpu() now uses resetpriority_thread() instead of just calling sched_prio() directly after it updates a kse group's user priority. - sched_clock() now uses resetpriority_thread() rather than writing directly to td_priority. - sched_nice() now updates all the priorities of the threads after the group priority has been adjusted. Discussed with: bde Reviewed by: ups, jeffr Tested on: 4bsd, ule Tested on: i386, alpha, sparc64
2004-12-30 20:52:44 +00:00
if (ke->ke_thread->td_flags & TDF_BORROWING) {
- In kseq_choose(), don't recalculate slice values for processes with a nice of 0. Doing so can cause an infinite loop because they should be running, but a nice -20 process could prevent them from doing so. - Add a new flag KEF_PRIOELEV to flag a thread that has had its priority elevated due to priority propagation. If a thread has had its priority elevated, we assume that it must go on the current queue and it must get a slice. - In sched_userret() if our priority was elevated and we shouldn't have a timeslice, yield here until we should. Found/Tested by: glebius
2004-12-14 10:34:27 +00:00
ke->ke_slice = SCHED_SLICE_MIN;
return;
}
- Add in support for KSEs with 0 slice values on the run queue. If we try to select a KSE with a slice of 0 we will update its slice and insert it onto the next queue. - Pass the KSE instead of the ksegrp into sched_slice(). This more accurately reflects the behavior of the code. Slices are granted to kses. - Add a function kseq_nice_min() which finds the smallest nice value assigned to the kseg of any KSE on the queue. - Rewrite the logic in sched_slice(). Add a large comment describing the new slice selection scheme. To summarize, slices are assigned based on the nice value. Priorities are still calculated based on the nice and interactivity of a process. Slice sizes of 0 may be granted for KSEs whos nice is 20 or futher away from the lowest nice on the run queue. Other nice values are scaled across the range [min, min+20]. This fixes ULEs bad behavior with positively niced processes.
2003-04-02 06:46:43 +00:00
/*
* Rationale:
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
* KSEs in interactive ksegs get a minimal slice so that we
- Add in support for KSEs with 0 slice values on the run queue. If we try to select a KSE with a slice of 0 we will update its slice and insert it onto the next queue. - Pass the KSE instead of the ksegrp into sched_slice(). This more accurately reflects the behavior of the code. Slices are granted to kses. - Add a function kseq_nice_min() which finds the smallest nice value assigned to the kseg of any KSE on the queue. - Rewrite the logic in sched_slice(). Add a large comment describing the new slice selection scheme. To summarize, slices are assigned based on the nice value. Priorities are still calculated based on the nice and interactivity of a process. Slice sizes of 0 may be granted for KSEs whos nice is 20 or futher away from the lowest nice on the run queue. Other nice values are scaled across the range [min, min+20]. This fixes ULEs bad behavior with positively niced processes.
2003-04-02 06:46:43 +00:00
* quickly notice if it abuses its advantage.
*
* KSEs in non-interactive ksegs are assigned a slice that is
* based on the ksegs nice value relative to the least nice kseg
* on the run queue for this cpu.
*
* If the KSE is less nice than all others it gets the maximum
* slice and other KSEs will adjust their slice relative to
* this when they first expire.
*
* There is 20 point window that starts relative to the least
* nice kse on the run queue. Slice size is determined by
* the kse distance from the last nice ksegrp.
*
- Rename SCHED_PRI_NTHRESH to SCHED_SLICE_NTHRESH since it is only used in slice assignment. Add a comment describing what it does. - Remove a stale XXX comment, the nice should not impact the interactivity, nice adjustments only effect non-interactive tasks in ULE. - Don't allow nice -20 tasks to totally starve nice 0 tasks. Give them at least SCHED_SLICE_MIN ticks. We still allow nice 0 tasks to starve nice +20 tasks as intended.
2003-11-02 04:10:15 +00:00
* If the kse is outside of the window it will get no slice
* and will be reevaluated each time it is selected on the
* run queue. The exception to this is nice 0 ksegs when
* a nice -20 is running. They are always granted a minimum
* slice.
- Add in support for KSEs with 0 slice values on the run queue. If we try to select a KSE with a slice of 0 we will update its slice and insert it onto the next queue. - Pass the KSE instead of the ksegrp into sched_slice(). This more accurately reflects the behavior of the code. Slices are granted to kses. - Add a function kseq_nice_min() which finds the smallest nice value assigned to the kseg of any KSE on the queue. - Rewrite the logic in sched_slice(). Add a large comment describing the new slice selection scheme. To summarize, slices are assigned based on the nice value. Priorities are still calculated based on the nice and interactivity of a process. Slice sizes of 0 may be granted for KSEs whos nice is 20 or futher away from the lowest nice on the run queue. Other nice values are scaled across the range [min, min+20]. This fixes ULEs bad behavior with positively niced processes.
2003-04-02 06:46:43 +00:00
*/
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
if (!SCHED_INTERACTIVE(kg)) {
- Add in support for KSEs with 0 slice values on the run queue. If we try to select a KSE with a slice of 0 we will update its slice and insert it onto the next queue. - Pass the KSE instead of the ksegrp into sched_slice(). This more accurately reflects the behavior of the code. Slices are granted to kses. - Add a function kseq_nice_min() which finds the smallest nice value assigned to the kseg of any KSE on the queue. - Rewrite the logic in sched_slice(). Add a large comment describing the new slice selection scheme. To summarize, slices are assigned based on the nice value. Priorities are still calculated based on the nice and interactivity of a process. Slice sizes of 0 may be granted for KSEs whos nice is 20 or futher away from the lowest nice on the run queue. Other nice values are scaled across the range [min, min+20]. This fixes ULEs bad behavior with positively niced processes.
2003-04-02 06:46:43 +00:00
int nice;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
Nice, is a property of a process as a whole.. I mistakenly moved it to the ksegroup when breaking up the process structure. Put it back in the proc structure.
2004-06-16 00:26:31 +00:00
nice = kg->kg_proc->p_nice + (0 - kseq->ksq_nicemin);
- Remove the ksq_loads[] array. We are only interested in three counts, the total load, the timeshare load, and the number of threads that can be migrated to another cpu. Account for these seperately. - Introduce a KSE_CAN_MIGRATE() macro which determines whether or not a KSE can be migrated to another CPU. Currently, this only checks to see if we're an interrupt handler. Eventually this will also be used to support CPU binding.
2003-11-02 10:56:48 +00:00
if (kseq->ksq_load_timeshare == 0 ||
Nice, is a property of a process as a whole.. I mistakenly moved it to the ksegroup when breaking up the process structure. Put it back in the proc structure.
2004-06-16 00:26:31 +00:00
kg->kg_proc->p_nice < kseq->ksq_nicemin)
- Add in support for KSEs with 0 slice values on the run queue. If we try to select a KSE with a slice of 0 we will update its slice and insert it onto the next queue. - Pass the KSE instead of the ksegrp into sched_slice(). This more accurately reflects the behavior of the code. Slices are granted to kses. - Add a function kseq_nice_min() which finds the smallest nice value assigned to the kseg of any KSE on the queue. - Rewrite the logic in sched_slice(). Add a large comment describing the new slice selection scheme. To summarize, slices are assigned based on the nice value. Priorities are still calculated based on the nice and interactivity of a process. Slice sizes of 0 may be granted for KSEs whos nice is 20 or futher away from the lowest nice on the run queue. Other nice values are scaled across the range [min, min+20]. This fixes ULEs bad behavior with positively niced processes.
2003-04-02 06:46:43 +00:00
ke->ke_slice = SCHED_SLICE_MAX;
- Rename SCHED_PRI_NTHRESH to SCHED_SLICE_NTHRESH since it is only used in slice assignment. Add a comment describing what it does. - Remove a stale XXX comment, the nice should not impact the interactivity, nice adjustments only effect non-interactive tasks in ULE. - Don't allow nice -20 tasks to totally starve nice 0 tasks. Give them at least SCHED_SLICE_MIN ticks. We still allow nice 0 tasks to starve nice +20 tasks as intended.
2003-11-02 04:10:15 +00:00
else if (nice <= SCHED_SLICE_NTHRESH)
- Add in support for KSEs with 0 slice values on the run queue. If we try to select a KSE with a slice of 0 we will update its slice and insert it onto the next queue. - Pass the KSE instead of the ksegrp into sched_slice(). This more accurately reflects the behavior of the code. Slices are granted to kses. - Add a function kseq_nice_min() which finds the smallest nice value assigned to the kseg of any KSE on the queue. - Rewrite the logic in sched_slice(). Add a large comment describing the new slice selection scheme. To summarize, slices are assigned based on the nice value. Priorities are still calculated based on the nice and interactivity of a process. Slice sizes of 0 may be granted for KSEs whos nice is 20 or futher away from the lowest nice on the run queue. Other nice values are scaled across the range [min, min+20]. This fixes ULEs bad behavior with positively niced processes.
2003-04-02 06:46:43 +00:00
ke->ke_slice = SCHED_SLICE_NICE(nice);
Nice, is a property of a process as a whole.. I mistakenly moved it to the ksegroup when breaking up the process structure. Put it back in the proc structure.
2004-06-16 00:26:31 +00:00
else if (kg->kg_proc->p_nice == 0)
- Rename SCHED_PRI_NTHRESH to SCHED_SLICE_NTHRESH since it is only used in slice assignment. Add a comment describing what it does. - Remove a stale XXX comment, the nice should not impact the interactivity, nice adjustments only effect non-interactive tasks in ULE. - Don't allow nice -20 tasks to totally starve nice 0 tasks. Give them at least SCHED_SLICE_MIN ticks. We still allow nice 0 tasks to starve nice +20 tasks as intended.
2003-11-02 04:10:15 +00:00
ke->ke_slice = SCHED_SLICE_MIN;
- Add in support for KSEs with 0 slice values on the run queue. If we try to select a KSE with a slice of 0 we will update its slice and insert it onto the next queue. - Pass the KSE instead of the ksegrp into sched_slice(). This more accurately reflects the behavior of the code. Slices are granted to kses. - Add a function kseq_nice_min() which finds the smallest nice value assigned to the kseg of any KSE on the queue. - Rewrite the logic in sched_slice(). Add a large comment describing the new slice selection scheme. To summarize, slices are assigned based on the nice value. Priorities are still calculated based on the nice and interactivity of a process. Slice sizes of 0 may be granted for KSEs whos nice is 20 or futher away from the lowest nice on the run queue. Other nice values are scaled across the range [min, min+20]. This fixes ULEs bad behavior with positively niced processes.
2003-04-02 06:46:43 +00:00
else
ke->ke_slice = 0;
} else
- Running interactive tasks with the minimum time-slice is fine for vi and sh, but not so great for mozilla, X, etc. Add a fixed define for the slice size granted to interactive KSEs.
2003-12-20 12:54:35 +00:00
ke->ke_slice = SCHED_SLICE_INTERACTIVE;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
- Add in support for KSEs with 0 slice values on the run queue. If we try to select a KSE with a slice of 0 we will update its slice and insert it onto the next queue. - Pass the KSE instead of the ksegrp into sched_slice(). This more accurately reflects the behavior of the code. Slices are granted to kses. - Add a function kseq_nice_min() which finds the smallest nice value assigned to the kseg of any KSE on the queue. - Rewrite the logic in sched_slice(). Add a large comment describing the new slice selection scheme. To summarize, slices are assigned based on the nice value. Priorities are still calculated based on the nice and interactivity of a process. Slice sizes of 0 may be granted for KSEs whos nice is 20 or futher away from the lowest nice on the run queue. Other nice values are scaled across the range [min, min+20]. This fixes ULEs bad behavior with positively niced processes.
2003-04-02 06:46:43 +00:00
return;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
}
- Change sched_interact_update() to only accept slp+runtime values between 0 and SCHED_SLP_RUN_MAX * 2. This allows us to simplify the algorithm quite a bit. Before, it dealt with arbitrary values which required us to do nasty integer division tricks that didn't quite work out correctly. - Chnage sched_wakeup() to detect conditions where the slp+runtime could exceed SCHED_SLP_RUN_MAX * 2. This can happen if we go to sleep for longer than 6 seconds. In this case, we'll just clear the runtime and set the sleep time to the max. - Define a new function, sched_interact_fork() which updates the slp+runtime of a newly forked thread. We want to limit the amount of history retained from the parent so that we learn the child's behavior quickly. We don't, however want to decay it to nothing. Previously, we would simply divide each parameter by 100 whenever we forked. After a few forks the values would reach 0 and tasks would not be considered interactive. - Add another KTR entry, cleanup some existing entries. - Remove a useless sched_interact_update() from sched_priority(). This is already done by the callers that require it.
2003-11-02 03:36:33 +00:00
/*
* This routine enforces a maximum limit on the amount of scheduling history
* kept. It is called after either the slptime or runtime is adjusted.
* This routine will not operate correctly when slp or run times have been
* adjusted to more than double their maximum.
*/
- Add a new function "sched_interact_update()" that scales back the sleep and run time. - Scale the sleep and run time back via sched_interact_update() in more places. This is to keep the statistic more accurate. - Charge a parent one tick for forking a child. - Add only the run time and not the sleep time to the parents kg when a thread exits. This allows us to give a penalty for having an expensive thread exit but does not give a bonus for having an interactive thread exit. - Change the SLP_RUN_THROTTLE to limit us to 4/5th and not 1/2. - Change the SLP_RUN_MAX to two seconds. This keeps bursty interactive applications like mozilla and openoffice in the interactive range even through expensive tasks. - Recalculate the slice after every sleep. This ensures that once a task has been marked interactive it only has a slice of 1 at the risk of giving tasks that sleep for a very brief period a longer time slice.
2003-06-17 06:39:51 +00:00
static void
sched_interact_update(struct ksegrp *kg)
{
- Change sched_interact_update() to only accept slp+runtime values between 0 and SCHED_SLP_RUN_MAX * 2. This allows us to simplify the algorithm quite a bit. Before, it dealt with arbitrary values which required us to do nasty integer division tricks that didn't quite work out correctly. - Chnage sched_wakeup() to detect conditions where the slp+runtime could exceed SCHED_SLP_RUN_MAX * 2. This can happen if we go to sleep for longer than 6 seconds. In this case, we'll just clear the runtime and set the sleep time to the max. - Define a new function, sched_interact_fork() which updates the slp+runtime of a newly forked thread. We want to limit the amount of history retained from the parent so that we learn the child's behavior quickly. We don't, however want to decay it to nothing. Previously, we would simply divide each parameter by 100 whenever we forked. After a few forks the values would reach 0 and tasks would not be considered interactive. - Add another KTR entry, cleanup some existing entries. - Remove a useless sched_interact_update() from sched_priority(). This is already done by the callers that require it.
2003-11-02 03:36:33 +00:00
int sum;
sum = kg->kg_runtime + kg->kg_slptime;
if (sum < SCHED_SLP_RUN_MAX)
return;
/*
* If we have exceeded by more than 1/5th then the algorithm below
* will not bring us back into range. Dividing by two here forces
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
* us into the range of [4/5 * SCHED_INTERACT_MAX, SCHED_INTERACT_MAX]
- Change sched_interact_update() to only accept slp+runtime values between 0 and SCHED_SLP_RUN_MAX * 2. This allows us to simplify the algorithm quite a bit. Before, it dealt with arbitrary values which required us to do nasty integer division tricks that didn't quite work out correctly. - Chnage sched_wakeup() to detect conditions where the slp+runtime could exceed SCHED_SLP_RUN_MAX * 2. This can happen if we go to sleep for longer than 6 seconds. In this case, we'll just clear the runtime and set the sleep time to the max. - Define a new function, sched_interact_fork() which updates the slp+runtime of a newly forked thread. We want to limit the amount of history retained from the parent so that we learn the child's behavior quickly. We don't, however want to decay it to nothing. Previously, we would simply divide each parameter by 100 whenever we forked. After a few forks the values would reach 0 and tasks would not be considered interactive. - Add another KTR entry, cleanup some existing entries. - Remove a useless sched_interact_update() from sched_priority(). This is already done by the callers that require it.
2003-11-02 03:36:33 +00:00
*/
- Use the proper constant in sched_interact_update(). Previously, SCHED_INTERACT_MAX was used where SCHED_SLP_RUN_MAX was needed. This was causing the interactivity scaler to lose history at a more dramatic rate than intended.
2004-04-04 19:12:56 +00:00
if (sum > (SCHED_SLP_RUN_MAX / 5) * 6) {
- Change sched_interact_update() to only accept slp+runtime values between 0 and SCHED_SLP_RUN_MAX * 2. This allows us to simplify the algorithm quite a bit. Before, it dealt with arbitrary values which required us to do nasty integer division tricks that didn't quite work out correctly. - Chnage sched_wakeup() to detect conditions where the slp+runtime could exceed SCHED_SLP_RUN_MAX * 2. This can happen if we go to sleep for longer than 6 seconds. In this case, we'll just clear the runtime and set the sleep time to the max. - Define a new function, sched_interact_fork() which updates the slp+runtime of a newly forked thread. We want to limit the amount of history retained from the parent so that we learn the child's behavior quickly. We don't, however want to decay it to nothing. Previously, we would simply divide each parameter by 100 whenever we forked. After a few forks the values would reach 0 and tasks would not be considered interactive. - Add another KTR entry, cleanup some existing entries. - Remove a useless sched_interact_update() from sched_priority(). This is already done by the callers that require it.
2003-11-02 03:36:33 +00:00
kg->kg_runtime /= 2;
kg->kg_slptime /= 2;
return;
}
kg->kg_runtime = (kg->kg_runtime / 5) * 4;
kg->kg_slptime = (kg->kg_slptime / 5) * 4;
}
- Use a better algorithm in sched_pctcpu_update() Contributed by: Thomaswuerfl@gmx.de - In sched_prio(), adjust the run queue for threads which may need to move to the current queue due to priority propagation . - In sched_switch(), fix style bug introduced when the KSE support went in. Columns are 80 chars wide, not 90. - In sched_switch(), Fix the comparison in the idle case and explicitly re-initialize the runq in the not propagated case. - Remove dead code in sched_clock(). - In sched_clock(), If we're an IDLE class td set NEEDRESCHED so that threads that have become runnable will get a chance to. - In sched_runnable(), if we're not the IDLETD, we should not consider curthread when examining the load. This mimics the 4BSD behavior of returning 0 when the only runnable thread is running. - In sched_userret(), remove the code for setting NEEDRESCHED entirely. This is not necessary and is not implemented in 4BSD. - Use the correct comparison in sched_add() when checking to see if an idle prio task has had it's priority temporarily elevated.
2003-10-27 06:47:05 +00:00
- Change sched_interact_update() to only accept slp+runtime values between 0 and SCHED_SLP_RUN_MAX * 2. This allows us to simplify the algorithm quite a bit. Before, it dealt with arbitrary values which required us to do nasty integer division tricks that didn't quite work out correctly. - Chnage sched_wakeup() to detect conditions where the slp+runtime could exceed SCHED_SLP_RUN_MAX * 2. This can happen if we go to sleep for longer than 6 seconds. In this case, we'll just clear the runtime and set the sleep time to the max. - Define a new function, sched_interact_fork() which updates the slp+runtime of a newly forked thread. We want to limit the amount of history retained from the parent so that we learn the child's behavior quickly. We don't, however want to decay it to nothing. Previously, we would simply divide each parameter by 100 whenever we forked. After a few forks the values would reach 0 and tasks would not be considered interactive. - Add another KTR entry, cleanup some existing entries. - Remove a useless sched_interact_update() from sched_priority(). This is already done by the callers that require it.
2003-11-02 03:36:33 +00:00
static void
sched_interact_fork(struct ksegrp *kg)
{
int ratio;
int sum;
sum = kg->kg_runtime + kg->kg_slptime;
if (sum > SCHED_SLP_RUN_FORK) {
ratio = sum / SCHED_SLP_RUN_FORK;
kg->kg_runtime /= ratio;
kg->kg_slptime /= ratio;
- Add a new function "sched_interact_update()" that scales back the sleep and run time. - Scale the sleep and run time back via sched_interact_update() in more places. This is to keep the statistic more accurate. - Charge a parent one tick for forking a child. - Add only the run time and not the sleep time to the parents kg when a thread exits. This allows us to give a penalty for having an expensive thread exit but does not give a bonus for having an interactive thread exit. - Change the SLP_RUN_THROTTLE to limit us to 4/5th and not 1/2. - Change the SLP_RUN_MAX to two seconds. This keeps bursty interactive applications like mozilla and openoffice in the interactive range even through expensive tasks. - Recalculate the slice after every sleep. This ensures that once a task has been marked interactive it only has a slice of 1 at the risk of giving tasks that sleep for a very brief period a longer time slice.
2003-06-17 06:39:51 +00:00
}
}
- Create a function sched_interact_score() which decides on the interactivity of a kseg and assigns it a value of 0 through 100. - Use sched_interact_score() to determine the dynamic priority. - Define SCHED_CURR() in terms of sched_interact_score(). - Adjust the maximum slice back down to 100ms. - Remove redundant clearing of ke_runq in sched_wakeup() - Clean up #defines and comment them.
2003-03-04 02:45:59 +00:00
static int
sched_interact_score(struct ksegrp *kg)
{
- Fix the maximum slice value. I accidentally checked in a value of '2' which meant no process would run for longer than 20ms. - Slightly redo the interactivity scorer. It follows the same algorithm but in a slightly more correct way. Previously values above half were incorrect. - Lower the interactivity threshold to 20. It seems that in testing non- interactive tasks are hardly ever near there and expensive interactive tasks can sometimes surpass it. This area needs more testing. - Remove an unnecessary KTR. - Fix a case where an idle thread that had an elevated priority due to priority prop. would be placed back on the idle queue. - Delay setting NEEDRESCHED until userret() for threads that haad their priority elevated while in kernel. This gives us the same context switch optimization as SCHED_4BSD. - Limit the child's slice to 1 in sched_fork_kse() so we detect its behavior more quickly. - Inhert some of the run/slp time from the child in sched_exit_ksegrp(). - Redo some of the priority comparisons so they are more clear. - Throttle the frequency of sched_pctcpu_update() so that rounding errors do not make it invalid.
2003-06-15 02:18:29 +00:00
int div;
- Create a function sched_interact_score() which decides on the interactivity of a kseg and assigns it a value of 0 through 100. - Use sched_interact_score() to determine the dynamic priority. - Define SCHED_CURR() in terms of sched_interact_score(). - Adjust the maximum slice back down to 100ms. - Remove redundant clearing of ke_runq in sched_wakeup() - Clean up #defines and comment them.
2003-03-04 02:45:59 +00:00
if (kg->kg_runtime > kg->kg_slptime) {
- Fix the maximum slice value. I accidentally checked in a value of '2' which meant no process would run for longer than 20ms. - Slightly redo the interactivity scorer. It follows the same algorithm but in a slightly more correct way. Previously values above half were incorrect. - Lower the interactivity threshold to 20. It seems that in testing non- interactive tasks are hardly ever near there and expensive interactive tasks can sometimes surpass it. This area needs more testing. - Remove an unnecessary KTR. - Fix a case where an idle thread that had an elevated priority due to priority prop. would be placed back on the idle queue. - Delay setting NEEDRESCHED until userret() for threads that haad their priority elevated while in kernel. This gives us the same context switch optimization as SCHED_4BSD. - Limit the child's slice to 1 in sched_fork_kse() so we detect its behavior more quickly. - Inhert some of the run/slp time from the child in sched_exit_ksegrp(). - Redo some of the priority comparisons so they are more clear. - Throttle the frequency of sched_pctcpu_update() so that rounding errors do not make it invalid.
2003-06-15 02:18:29 +00:00
div = max(1, kg->kg_runtime / SCHED_INTERACT_HALF);
return (SCHED_INTERACT_HALF +
(SCHED_INTERACT_HALF - (kg->kg_slptime / div)));
} if (kg->kg_slptime > kg->kg_runtime) {
div = max(1, kg->kg_slptime / SCHED_INTERACT_HALF);
return (kg->kg_runtime / div);
- Create a function sched_interact_score() which decides on the interactivity of a kseg and assigns it a value of 0 through 100. - Use sched_interact_score() to determine the dynamic priority. - Define SCHED_CURR() in terms of sched_interact_score(). - Adjust the maximum slice back down to 100ms. - Remove redundant clearing of ke_runq in sched_wakeup() - Clean up #defines and comment them.
2003-03-04 02:45:59 +00:00
}
- Fix the maximum slice value. I accidentally checked in a value of '2' which meant no process would run for longer than 20ms. - Slightly redo the interactivity scorer. It follows the same algorithm but in a slightly more correct way. Previously values above half were incorrect. - Lower the interactivity threshold to 20. It seems that in testing non- interactive tasks are hardly ever near there and expensive interactive tasks can sometimes surpass it. This area needs more testing. - Remove an unnecessary KTR. - Fix a case where an idle thread that had an elevated priority due to priority prop. would be placed back on the idle queue. - Delay setting NEEDRESCHED until userret() for threads that haad their priority elevated while in kernel. This gives us the same context switch optimization as SCHED_4BSD. - Limit the child's slice to 1 in sched_fork_kse() so we detect its behavior more quickly. - Inhert some of the run/slp time from the child in sched_exit_ksegrp(). - Redo some of the priority comparisons so they are more clear. - Throttle the frequency of sched_pctcpu_update() so that rounding errors do not make it invalid.
2003-06-15 02:18:29 +00:00
/*
* This can happen if slptime and runtime are 0.
*/
return (0);
- Create a function sched_interact_score() which decides on the interactivity of a kseg and assigns it a value of 0 through 100. - Use sched_interact_score() to determine the dynamic priority. - Define SCHED_CURR() in terms of sched_interact_score(). - Adjust the maximum slice back down to 100ms. - Remove redundant clearing of ke_runq in sched_wakeup() - Clean up #defines and comment them.
2003-03-04 02:45:59 +00:00
}
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
/*
* Very early in the boot some setup of scheduler-specific
* parts of proc0 and of soem scheduler resources needs to be done.
* Called from:
* proc0_init()
*/
void
schedinit(void)
{
/*
* Set up the scheduler specific parts of proc0.
*/
proc0.p_sched = NULL; /* XXX */
Use some macros to trach available scheduler slots to allow easier debugging. MFC after: 4 days
2004-10-05 21:10:44 +00:00
ksegrp0.kg_sched = &kg_sched0;
thread0.td_sched = &kse0;
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
kse0.ke_thread = &thread0;
kse0.ke_state = KES_THREAD;
kg_sched0.skg_concurrency = 1;
kg_sched0.skg_avail_opennings = 0; /* we are already running */
}
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
/*
* This is only somewhat accurate since given many processes of the same
* priority they will switch when their slices run out, which will be
* at most SCHED_SLICE_MAX.
*/
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
int
sched_rr_interval(void)
{
return (SCHED_SLICE_MAX);
}
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
static void
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
sched_pctcpu_update(struct kse *ke)
{
/*
* Adjust counters and watermark for pctcpu calc.
- Fix the maximum slice value. I accidentally checked in a value of '2' which meant no process would run for longer than 20ms. - Slightly redo the interactivity scorer. It follows the same algorithm but in a slightly more correct way. Previously values above half were incorrect. - Lower the interactivity threshold to 20. It seems that in testing non- interactive tasks are hardly ever near there and expensive interactive tasks can sometimes surpass it. This area needs more testing. - Remove an unnecessary KTR. - Fix a case where an idle thread that had an elevated priority due to priority prop. would be placed back on the idle queue. - Delay setting NEEDRESCHED until userret() for threads that haad their priority elevated while in kernel. This gives us the same context switch optimization as SCHED_4BSD. - Limit the child's slice to 1 in sched_fork_kse() so we detect its behavior more quickly. - Inhert some of the run/slp time from the child in sched_exit_ksegrp(). - Redo some of the priority comparisons so they are more clear. - Throttle the frequency of sched_pctcpu_update() so that rounding errors do not make it invalid.
2003-06-15 02:18:29 +00:00
*/
- Somewhere along the line I stupidly removed critical logic from sched_ptcpu_update(). This caused erroneous cpu times in TOP for processes that were asleep. Replace the code that was removed.
2003-09-20 02:05:58 +00:00
if (ke->ke_ltick > ticks - SCHED_CPU_TICKS) {
/*
* Shift the tick count out so that the divide doesn't
* round away our results.
*/
ke->ke_ticks <<= 10;
ke->ke_ticks = (ke->ke_ticks / (ticks - ke->ke_ftick)) *
SCHED_CPU_TICKS;
ke->ke_ticks >>= 10;
} else
ke->ke_ticks = 0;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
ke->ke_ltick = ticks;
ke->ke_ftick = ke->ke_ltick - SCHED_CPU_TICKS;
}
void
Rework the interface between priority propagation (lending) and the schedulers a bit to ensure more correct handling of priorities and fewer priority inversions: - Add two functions to the sched(9) API to handle priority lending: sched_lend_prio() and sched_unlend_prio(). The turnstile code uses these functions to ask the scheduler to lend a thread a set priority and to tell the scheduler when it thinks it is ok for a thread to stop borrowing priority. The unlend case is slightly complex in that the turnstile code tells the scheduler what the minimum priority of the thread needs to be to satisfy the requirements of any other threads blocked on locks owned by the thread in question. The scheduler then decides where the thread can go back to normal mode (if it's normal priority is high enough to satisfy the pending lock requests) or it it should continue to use the priority specified to the sched_unlend_prio() call. This involves adding a new per-thread flag TDF_BORROWING that replaces the ULE-only kse flag for priority elevation. - Schedulers now refuse to lower the priority of a thread that is currently borrowing another therad's priority. - If a scheduler changes the priority of a thread that is currently sitting on a turnstile, it will call a new function turnstile_adjust() to inform the turnstile code of the change. This function resorts the thread on the priority list of the turnstile if needed, and if the thread ends up at the head of the list (due to having the highest priority) and its priority was raised, then it will propagate that new priority to the owner of the lock it is blocked on. Some additional fixes specific to the 4BSD scheduler include: - Common code for updating the priority of a thread when the user priority of its associated kse group has been consolidated in a new static function resetpriority_thread(). One change to this function is that it will now only adjust the priority of a thread if it already has a time sharing priority, thus preserving any boosts from a tsleep() until the thread returns to userland. Also, resetpriority() no longer calls maybe_resched() on each thread in the group. Instead, the code calling resetpriority() is responsible for calling resetpriority_thread() on any threads that need to be updated. - schedcpu() now uses resetpriority_thread() instead of just calling sched_prio() directly after it updates a kse group's user priority. - sched_clock() now uses resetpriority_thread() rather than writing directly to td_priority. - sched_nice() now updates all the priorities of the threads after the group priority has been adjusted. Discussed with: bde Reviewed by: ups, jeffr Tested on: 4bsd, ule Tested on: i386, alpha, sparc64
2004-12-30 20:52:44 +00:00
sched_thread_priority(struct thread *td, u_char prio)
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
{
- Use a better algorithm in sched_pctcpu_update() Contributed by: Thomaswuerfl@gmx.de - In sched_prio(), adjust the run queue for threads which may need to move to the current queue due to priority propagation . - In sched_switch(), fix style bug introduced when the KSE support went in. Columns are 80 chars wide, not 90. - In sched_switch(), Fix the comparison in the idle case and explicitly re-initialize the runq in the not propagated case. - Remove dead code in sched_clock(). - In sched_clock(), If we're an IDLE class td set NEEDRESCHED so that threads that have become runnable will get a chance to. - In sched_runnable(), if we're not the IDLETD, we should not consider curthread when examining the load. This mimics the 4BSD behavior of returning 0 when the only runnable thread is running. - In sched_userret(), remove the code for setting NEEDRESCHED entirely. This is not necessary and is not implemented in 4BSD. - Use the correct comparison in sched_add() when checking to see if an idle prio task has had it's priority temporarily elevated.
2003-10-27 06:47:05 +00:00
struct kse *ke;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
- Remove earlier KTR_ULE tracepoints. - Define new KTR_SCHED points so that we can graph the operation of the scheduler.
2004-12-26 00:15:33 +00:00
CTR6(KTR_SCHED, "sched_prio: %p(%s) prio %d newprio %d by %p(%s)",
td, td->td_proc->p_comm, td->td_priority, prio, curthread,
curthread->td_proc->p_comm);
- Use a better algorithm in sched_pctcpu_update() Contributed by: Thomaswuerfl@gmx.de - In sched_prio(), adjust the run queue for threads which may need to move to the current queue due to priority propagation . - In sched_switch(), fix style bug introduced when the KSE support went in. Columns are 80 chars wide, not 90. - In sched_switch(), Fix the comparison in the idle case and explicitly re-initialize the runq in the not propagated case. - Remove dead code in sched_clock(). - In sched_clock(), If we're an IDLE class td set NEEDRESCHED so that threads that have become runnable will get a chance to. - In sched_runnable(), if we're not the IDLETD, we should not consider curthread when examining the load. This mimics the 4BSD behavior of returning 0 when the only runnable thread is running. - In sched_userret(), remove the code for setting NEEDRESCHED entirely. This is not necessary and is not implemented in 4BSD. - Use the correct comparison in sched_add() when checking to see if an idle prio task has had it's priority temporarily elevated.
2003-10-27 06:47:05 +00:00
ke = td->td_kse;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
mtx_assert(&sched_lock, MA_OWNED);
Rework the interface between priority propagation (lending) and the schedulers a bit to ensure more correct handling of priorities and fewer priority inversions: - Add two functions to the sched(9) API to handle priority lending: sched_lend_prio() and sched_unlend_prio(). The turnstile code uses these functions to ask the scheduler to lend a thread a set priority and to tell the scheduler when it thinks it is ok for a thread to stop borrowing priority. The unlend case is slightly complex in that the turnstile code tells the scheduler what the minimum priority of the thread needs to be to satisfy the requirements of any other threads blocked on locks owned by the thread in question. The scheduler then decides where the thread can go back to normal mode (if it's normal priority is high enough to satisfy the pending lock requests) or it it should continue to use the priority specified to the sched_unlend_prio() call. This involves adding a new per-thread flag TDF_BORROWING that replaces the ULE-only kse flag for priority elevation. - Schedulers now refuse to lower the priority of a thread that is currently borrowing another therad's priority. - If a scheduler changes the priority of a thread that is currently sitting on a turnstile, it will call a new function turnstile_adjust() to inform the turnstile code of the change. This function resorts the thread on the priority list of the turnstile if needed, and if the thread ends up at the head of the list (due to having the highest priority) and its priority was raised, then it will propagate that new priority to the owner of the lock it is blocked on. Some additional fixes specific to the 4BSD scheduler include: - Common code for updating the priority of a thread when the user priority of its associated kse group has been consolidated in a new static function resetpriority_thread(). One change to this function is that it will now only adjust the priority of a thread if it already has a time sharing priority, thus preserving any boosts from a tsleep() until the thread returns to userland. Also, resetpriority() no longer calls maybe_resched() on each thread in the group. Instead, the code calling resetpriority() is responsible for calling resetpriority_thread() on any threads that need to be updated. - schedcpu() now uses resetpriority_thread() instead of just calling sched_prio() directly after it updates a kse group's user priority. - sched_clock() now uses resetpriority_thread() rather than writing directly to td_priority. - sched_nice() now updates all the priorities of the threads after the group priority has been adjusted. Discussed with: bde Reviewed by: ups, jeffr Tested on: 4bsd, ule Tested on: i386, alpha, sparc64
2004-12-30 20:52:44 +00:00
if (td->td_priority == prio)
return;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
if (TD_ON_RUNQ(td)) {
- Use a better algorithm in sched_pctcpu_update() Contributed by: Thomaswuerfl@gmx.de - In sched_prio(), adjust the run queue for threads which may need to move to the current queue due to priority propagation . - In sched_switch(), fix style bug introduced when the KSE support went in. Columns are 80 chars wide, not 90. - In sched_switch(), Fix the comparison in the idle case and explicitly re-initialize the runq in the not propagated case. - Remove dead code in sched_clock(). - In sched_clock(), If we're an IDLE class td set NEEDRESCHED so that threads that have become runnable will get a chance to. - In sched_runnable(), if we're not the IDLETD, we should not consider curthread when examining the load. This mimics the 4BSD behavior of returning 0 when the only runnable thread is running. - In sched_userret(), remove the code for setting NEEDRESCHED entirely. This is not necessary and is not implemented in 4BSD. - Use the correct comparison in sched_add() when checking to see if an idle prio task has had it's priority temporarily elevated.
2003-10-27 06:47:05 +00:00
/*
* If the priority has been elevated due to priority
* propagation, we may have to move ourselves to a new
* queue. We still call adjustrunqueue below in case kse
* needs to fix things up.
*/
- In kseq_choose(), don't recalculate slice values for processes with a nice of 0. Doing so can cause an infinite loop because they should be running, but a nice -20 process could prevent them from doing so. - Add a new flag KEF_PRIOELEV to flag a thread that has had its priority elevated due to priority propagation. If a thread has had its priority elevated, we assume that it must go on the current queue and it must get a slice. - In sched_userret() if our priority was elevated and we shouldn't have a timeslice, yield here until we should. Found/Tested by: glebius
2004-12-14 10:34:27 +00:00
if (prio < td->td_priority && ke->ke_runq != NULL &&
- In sched_prio() only force us onto the current queue if our priority is being elevated (numerically smaller).
2003-11-02 04:25:59 +00:00
(ke->ke_flags & KEF_ASSIGNED) == 0 &&
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
ke->ke_runq != KSEQ_CPU(ke->ke_cpu)->ksq_curr) {
- Use a better algorithm in sched_pctcpu_update() Contributed by: Thomaswuerfl@gmx.de - In sched_prio(), adjust the run queue for threads which may need to move to the current queue due to priority propagation . - In sched_switch(), fix style bug introduced when the KSE support went in. Columns are 80 chars wide, not 90. - In sched_switch(), Fix the comparison in the idle case and explicitly re-initialize the runq in the not propagated case. - Remove dead code in sched_clock(). - In sched_clock(), If we're an IDLE class td set NEEDRESCHED so that threads that have become runnable will get a chance to. - In sched_runnable(), if we're not the IDLETD, we should not consider curthread when examining the load. This mimics the 4BSD behavior of returning 0 when the only runnable thread is running. - In sched_userret(), remove the code for setting NEEDRESCHED entirely. This is not necessary and is not implemented in 4BSD. - Use the correct comparison in sched_add() when checking to see if an idle prio task has had it's priority temporarily elevated.
2003-10-27 06:47:05 +00:00
runq_remove(ke->ke_runq, ke);
ke->ke_runq = KSEQ_CPU(ke->ke_cpu)->ksq_curr;
When preempting a thread, put it back on the HEAD of its run queue. (Only really implemented in 4bsd) MFC after: 4 days
2004-10-05 22:03:10 +00:00
runq_add(ke->ke_runq, ke, 0);
- Use a better algorithm in sched_pctcpu_update() Contributed by: Thomaswuerfl@gmx.de - In sched_prio(), adjust the run queue for threads which may need to move to the current queue due to priority propagation . - In sched_switch(), fix style bug introduced when the KSE support went in. Columns are 80 chars wide, not 90. - In sched_switch(), Fix the comparison in the idle case and explicitly re-initialize the runq in the not propagated case. - Remove dead code in sched_clock(). - In sched_clock(), If we're an IDLE class td set NEEDRESCHED so that threads that have become runnable will get a chance to. - In sched_runnable(), if we're not the IDLETD, we should not consider curthread when examining the load. This mimics the 4BSD behavior of returning 0 when the only runnable thread is running. - In sched_userret(), remove the code for setting NEEDRESCHED entirely. This is not necessary and is not implemented in 4BSD. - Use the correct comparison in sched_add() when checking to see if an idle prio task has had it's priority temporarily elevated.
2003-10-27 06:47:05 +00:00
}
- Introduce a new flag KEF_HOLD that prevents sched_add() from doing a migration. Use this in sched_prio() and sched_switch() to stop us from migrating threads that are in short term sleeps or are runnable. These extra migrations were added in the patches to support KSE. - Only set NEEDRESCHED if the thread we're adding in sched_add() is a lower priority and is being placed on the current queue. - Fix some minor whitespace problems.
2004-08-12 07:56:33 +00:00
/*
* Hold this kse on this cpu so that sched_prio() doesn't
* cause excessive migration. We only want migration to
* happen as the result of a wakeup.
*/
ke->ke_flags |= KEF_HOLD;
Let SA process work under ULE scheduler, originally it would panic kernel. Reviewed by: jeff
2003-08-26 11:33:15 +00:00
adjustrunqueue(td, prio);
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
ke->ke_flags &= ~KEF_HOLD;
- Use a better algorithm in sched_pctcpu_update() Contributed by: Thomaswuerfl@gmx.de - In sched_prio(), adjust the run queue for threads which may need to move to the current queue due to priority propagation . - In sched_switch(), fix style bug introduced when the KSE support went in. Columns are 80 chars wide, not 90. - In sched_switch(), Fix the comparison in the idle case and explicitly re-initialize the runq in the not propagated case. - Remove dead code in sched_clock(). - In sched_clock(), If we're an IDLE class td set NEEDRESCHED so that threads that have become runnable will get a chance to. - In sched_runnable(), if we're not the IDLETD, we should not consider curthread when examining the load. This mimics the 4BSD behavior of returning 0 when the only runnable thread is running. - In sched_userret(), remove the code for setting NEEDRESCHED entirely. This is not necessary and is not implemented in 4BSD. - Use the correct comparison in sched_add() when checking to see if an idle prio task has had it's priority temporarily elevated.
2003-10-27 06:47:05 +00:00
} else
Let SA process work under ULE scheduler, originally it would panic kernel. Reviewed by: jeff
2003-08-26 11:33:15 +00:00
td->td_priority = prio;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
}
Rework the interface between priority propagation (lending) and the schedulers a bit to ensure more correct handling of priorities and fewer priority inversions: - Add two functions to the sched(9) API to handle priority lending: sched_lend_prio() and sched_unlend_prio(). The turnstile code uses these functions to ask the scheduler to lend a thread a set priority and to tell the scheduler when it thinks it is ok for a thread to stop borrowing priority. The unlend case is slightly complex in that the turnstile code tells the scheduler what the minimum priority of the thread needs to be to satisfy the requirements of any other threads blocked on locks owned by the thread in question. The scheduler then decides where the thread can go back to normal mode (if it's normal priority is high enough to satisfy the pending lock requests) or it it should continue to use the priority specified to the sched_unlend_prio() call. This involves adding a new per-thread flag TDF_BORROWING that replaces the ULE-only kse flag for priority elevation. - Schedulers now refuse to lower the priority of a thread that is currently borrowing another therad's priority. - If a scheduler changes the priority of a thread that is currently sitting on a turnstile, it will call a new function turnstile_adjust() to inform the turnstile code of the change. This function resorts the thread on the priority list of the turnstile if needed, and if the thread ends up at the head of the list (due to having the highest priority) and its priority was raised, then it will propagate that new priority to the owner of the lock it is blocked on. Some additional fixes specific to the 4BSD scheduler include: - Common code for updating the priority of a thread when the user priority of its associated kse group has been consolidated in a new static function resetpriority_thread(). One change to this function is that it will now only adjust the priority of a thread if it already has a time sharing priority, thus preserving any boosts from a tsleep() until the thread returns to userland. Also, resetpriority() no longer calls maybe_resched() on each thread in the group. Instead, the code calling resetpriority() is responsible for calling resetpriority_thread() on any threads that need to be updated. - schedcpu() now uses resetpriority_thread() instead of just calling sched_prio() directly after it updates a kse group's user priority. - sched_clock() now uses resetpriority_thread() rather than writing directly to td_priority. - sched_nice() now updates all the priorities of the threads after the group priority has been adjusted. Discussed with: bde Reviewed by: ups, jeffr Tested on: 4bsd, ule Tested on: i386, alpha, sparc64
2004-12-30 20:52:44 +00:00
/*
* Update a thread's priority when it is lent another thread's
* priority.
*/
void
sched_lend_prio(struct thread *td, u_char prio)
{
td->td_flags |= TDF_BORROWING;
sched_thread_priority(td, prio);
}
/*
* Restore a thread's priority when priority propagation is
* over. The prio argument is the minimum priority the thread
* needs to have to satisfy other possible priority lending
* requests. If the thread's regular priority is less
* important than prio, the thread will keep a priority boost
* of prio.
*/
void
sched_unlend_prio(struct thread *td, u_char prio)
{
u_char base_pri;
if (td->td_base_pri >= PRI_MIN_TIMESHARE &&
td->td_base_pri <= PRI_MAX_TIMESHARE)
base_pri = td->td_ksegrp->kg_user_pri;
else
base_pri = td->td_base_pri;
if (prio >= base_pri) {
Fix a typo and two whitespace nits.
2004-12-30 22:17:00 +00:00
td->td_flags &= ~TDF_BORROWING;
Rework the interface between priority propagation (lending) and the schedulers a bit to ensure more correct handling of priorities and fewer priority inversions: - Add two functions to the sched(9) API to handle priority lending: sched_lend_prio() and sched_unlend_prio(). The turnstile code uses these functions to ask the scheduler to lend a thread a set priority and to tell the scheduler when it thinks it is ok for a thread to stop borrowing priority. The unlend case is slightly complex in that the turnstile code tells the scheduler what the minimum priority of the thread needs to be to satisfy the requirements of any other threads blocked on locks owned by the thread in question. The scheduler then decides where the thread can go back to normal mode (if it's normal priority is high enough to satisfy the pending lock requests) or it it should continue to use the priority specified to the sched_unlend_prio() call. This involves adding a new per-thread flag TDF_BORROWING that replaces the ULE-only kse flag for priority elevation. - Schedulers now refuse to lower the priority of a thread that is currently borrowing another therad's priority. - If a scheduler changes the priority of a thread that is currently sitting on a turnstile, it will call a new function turnstile_adjust() to inform the turnstile code of the change. This function resorts the thread on the priority list of the turnstile if needed, and if the thread ends up at the head of the list (due to having the highest priority) and its priority was raised, then it will propagate that new priority to the owner of the lock it is blocked on. Some additional fixes specific to the 4BSD scheduler include: - Common code for updating the priority of a thread when the user priority of its associated kse group has been consolidated in a new static function resetpriority_thread(). One change to this function is that it will now only adjust the priority of a thread if it already has a time sharing priority, thus preserving any boosts from a tsleep() until the thread returns to userland. Also, resetpriority() no longer calls maybe_resched() on each thread in the group. Instead, the code calling resetpriority() is responsible for calling resetpriority_thread() on any threads that need to be updated. - schedcpu() now uses resetpriority_thread() instead of just calling sched_prio() directly after it updates a kse group's user priority. - sched_clock() now uses resetpriority_thread() rather than writing directly to td_priority. - sched_nice() now updates all the priorities of the threads after the group priority has been adjusted. Discussed with: bde Reviewed by: ups, jeffr Tested on: 4bsd, ule Tested on: i386, alpha, sparc64
2004-12-30 20:52:44 +00:00
sched_thread_priority(td, base_pri);
} else
sched_lend_prio(td, prio);
}
void
sched_prio(struct thread *td, u_char prio)
{
u_char oldprio;
/* First, update the base priority. */
td->td_base_pri = prio;
/*
Fix a typo and two whitespace nits.
2004-12-30 22:17:00 +00:00
* If the thread is borrowing another thread's priority, don't
Rework the interface between priority propagation (lending) and the schedulers a bit to ensure more correct handling of priorities and fewer priority inversions: - Add two functions to the sched(9) API to handle priority lending: sched_lend_prio() and sched_unlend_prio(). The turnstile code uses these functions to ask the scheduler to lend a thread a set priority and to tell the scheduler when it thinks it is ok for a thread to stop borrowing priority. The unlend case is slightly complex in that the turnstile code tells the scheduler what the minimum priority of the thread needs to be to satisfy the requirements of any other threads blocked on locks owned by the thread in question. The scheduler then decides where the thread can go back to normal mode (if it's normal priority is high enough to satisfy the pending lock requests) or it it should continue to use the priority specified to the sched_unlend_prio() call. This involves adding a new per-thread flag TDF_BORROWING that replaces the ULE-only kse flag for priority elevation. - Schedulers now refuse to lower the priority of a thread that is currently borrowing another therad's priority. - If a scheduler changes the priority of a thread that is currently sitting on a turnstile, it will call a new function turnstile_adjust() to inform the turnstile code of the change. This function resorts the thread on the priority list of the turnstile if needed, and if the thread ends up at the head of the list (due to having the highest priority) and its priority was raised, then it will propagate that new priority to the owner of the lock it is blocked on. Some additional fixes specific to the 4BSD scheduler include: - Common code for updating the priority of a thread when the user priority of its associated kse group has been consolidated in a new static function resetpriority_thread(). One change to this function is that it will now only adjust the priority of a thread if it already has a time sharing priority, thus preserving any boosts from a tsleep() until the thread returns to userland. Also, resetpriority() no longer calls maybe_resched() on each thread in the group. Instead, the code calling resetpriority() is responsible for calling resetpriority_thread() on any threads that need to be updated. - schedcpu() now uses resetpriority_thread() instead of just calling sched_prio() directly after it updates a kse group's user priority. - sched_clock() now uses resetpriority_thread() rather than writing directly to td_priority. - sched_nice() now updates all the priorities of the threads after the group priority has been adjusted. Discussed with: bde Reviewed by: ups, jeffr Tested on: 4bsd, ule Tested on: i386, alpha, sparc64
2004-12-30 20:52:44 +00:00
* ever lower the priority.
*/
if (td->td_flags & TDF_BORROWING && td->td_priority < prio)
return;
/* Change the real priority. */
oldprio = td->td_priority;
sched_thread_priority(td, prio);
/*
* If the thread is on a turnstile, then let the turnstile update
* its state.
*/
if (TD_ON_LOCK(td) && oldprio != prio)
turnstile_adjust(td, oldprio);
}
Fix a typo and two whitespace nits.
2004-12-30 22:17:00 +00:00
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
void
Add some code to allow threads to nominat a sibling to run if theyu are going to sleep. MFC after: 1 week
2004-09-10 21:04:38 +00:00
sched_switch(struct thread *td, struct thread *newtd, int flags)
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
{
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
struct kseq *ksq;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
struct kse *ke;
mtx_assert(&sched_lock, MA_OWNED);
ke = td->td_kse;
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
ksq = KSEQ_SELF();
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
- Introduce a new flag KEF_HOLD that prevents sched_add() from doing a migration. Use this in sched_prio() and sched_switch() to stop us from migrating threads that are in short term sleeps or are runnable. These extra migrations were added in the patches to support KSE. - Only set NEEDRESCHED if the thread we're adding in sched_add() is a lower priority and is being placed on the current queue. - Fix some minor whitespace problems.
2004-08-12 07:56:33 +00:00
td->td_lastcpu = td->td_oncpu;
Move the _oncpu entry from the KSE to the thread. The entry in the KSE still exists but it's purpose will change a bit when we add the ability to lock a KSE to a cpu.
2003-04-10 17:35:44 +00:00
td->td_oncpu = NOCPU;
- Move TDF_OWEPREEMPT, TDF_OWEUPC, and TDF_USTATCLOCK over to td_pflags since they are only accessed by curthread and thus do not need any locking. - Move pr_addr and pr_ticks out of struct uprof (which is per-process) and directly into struct thread as td_profil_addr and td_profil_ticks as these variables are really per-thread. (They are used to defer an addupc_intr() that was too "hard" until ast()).
2004-07-16 21:04:55 +00:00
td->td_flags &= ~TDF_NEEDRESCHED;
td->td_pflags &= ~TDP_OWEPREEMPT;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
- In sched_switch(), if a thread has been assigned, don't touch the runqueues or load. These things have already been taken care of in sched_bind() which should be the only place that we're switching in an assigned thread.
2003-12-11 04:00:49 +00:00
/*
* If the KSE has been assigned it may be in the process of switching
* to the new cpu. This is the case in sched_bind().
*/
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
if (td == PCPU_GET(idlethread)) {
TD_SET_CAN_RUN(td);
} else if ((ke->ke_flags & KEF_ASSIGNED) == 0) {
/* We are ending our run so make our slot available again */
SLOT_RELEASE(td->td_ksegrp);
kseq_load_rem(ksq, ke);
if (TD_IS_RUNNING(td)) {
/*
* Don't allow the thread to migrate
* from a preemption.
*/
ke->ke_flags |= KEF_HOLD;
setrunqueue(td, (flags & SW_PREEMPT) ?
SRQ_OURSELF|SRQ_YIELDING|SRQ_PREEMPTED :
SRQ_OURSELF|SRQ_YIELDING);
ke->ke_flags &= ~KEF_HOLD;
} else if ((td->td_proc->p_flag & P_HADTHREADS) &&
(newtd == NULL || newtd->td_ksegrp != td->td_ksegrp))
/*
* We will not be on the run queue.
* So we must be sleeping or similar.
* Don't use the slot if we will need it
* for newtd.
*/
slot_fill(td->td_ksegrp);
- Only kse_reassign() in the !running case. Reported by: kris
2003-10-16 20:32:57 +00:00
}
Use some macros to trach available scheduler slots to allow easier debugging. MFC after: 4 days
2004-10-05 21:10:44 +00:00
if (newtd != NULL) {
When preempting a thread, put it back on the HEAD of its run queue. (Only really implemented in 4bsd) MFC after: 4 days
2004-10-05 22:03:10 +00:00
/*
* If we bring in a thread,
* then account for it as if it had been added to the
* run queue and then chosen.
*/
Oops. left out part of the diff. MFC after: 4 days
2004-10-05 21:26:27 +00:00
newtd->td_kse->ke_flags |= KEF_DIDRUN;
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
newtd->td_kse->ke_runq = ksq->ksq_curr;
Use some macros to trach available scheduler slots to allow easier debugging. MFC after: 4 days
2004-10-05 21:10:44 +00:00
SLOT_USE(newtd->td_ksegrp);
Fix whitespace botch that only showed up in the commit message diff :-/ MFC after: 4 days
2004-10-05 22:14:02 +00:00
TD_SET_RUNNING(newtd);
- Change mi_switch() and sched_switch() to accept an optional thread to switch to. If a non-NULL thread pointer is passed in, then the CPU will switch to that thread directly rather than calling choosethread() to pick a thread to choose to. - Make sched_switch() aware of idle threads and know to do TD_SET_CAN_RUN() instead of sticking them on the run queue rather than requiring all callers of mi_switch() to know to do this if they can be called from an idlethread. - Move constants for arguments to mi_switch() and thread_single() out of the middle of the function prototypes and up above into their own section.
2004-07-02 19:09:50 +00:00
kseq_load_add(KSEQ_SELF(), newtd->td_kse);
Use some macros to trach available scheduler slots to allow easier debugging. MFC after: 4 days
2004-10-05 21:10:44 +00:00
} else
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
newtd = choosethread();
- Collapse sched_switchin() and sched_switchout() into sched_switch(). Now mi_switch() calls sched_switch() which calls cpu_switch(). This is actually one less function call than it had been.
2003-10-16 08:53:46 +00:00
if (td != newtd)
cpu_switch(td, newtd);
sched_lock.mtx_lock = (uintptr_t)td;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
Move the _oncpu entry from the KSE to the thread. The entry in the KSE still exists but it's purpose will change a bit when we add the ability to lock a KSE to a cpu.
2003-04-10 17:35:44 +00:00
td->td_oncpu = PCPU_GET(cpuid);
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
}
void
Nice, is a property of a process as a whole.. I mistakenly moved it to the ksegroup when breaking up the process structure. Put it back in the proc structure.
2004-06-16 00:26:31 +00:00
sched_nice(struct proc *p, int nice)
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
{
Nice, is a property of a process as a whole.. I mistakenly moved it to the ksegroup when breaking up the process structure. Put it back in the proc structure.
2004-06-16 00:26:31 +00:00
struct ksegrp *kg;
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
struct kse *ke;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
struct thread *td;
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
struct kseq *kseq;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
Nice, is a property of a process as a whole.. I mistakenly moved it to the ksegroup when breaking up the process structure. Put it back in the proc structure.
2004-06-16 00:26:31 +00:00
PROC_LOCK_ASSERT(p, MA_OWNED);
- Assert that the proc lock and sched_lock are held in sched_nice(). - For the 4BSD scheduler, this means that all callers of the static function resetpriority() now always hold sched_lock, so don't lock sched_lock explicitly in that function.
2003-04-22 20:50:38 +00:00
mtx_assert(&sched_lock, MA_OWNED);
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
/*
* We need to adjust the nice counts for running KSEs.
*/
Nice, is a property of a process as a whole.. I mistakenly moved it to the ksegroup when breaking up the process structure. Put it back in the proc structure.
2004-06-16 00:26:31 +00:00
FOREACH_KSEGRP_IN_PROC(p, kg) {
if (kg->kg_pri_class == PRI_TIMESHARE) {
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
FOREACH_THREAD_IN_GROUP(kg, td) {
ke = td->td_kse;
Nice, is a property of a process as a whole.. I mistakenly moved it to the ksegroup when breaking up the process structure. Put it back in the proc structure.
2004-06-16 00:26:31 +00:00
if (ke->ke_runq == NULL)
continue;
kseq = KSEQ_CPU(ke->ke_cpu);
kseq_nice_rem(kseq, p->p_nice);
kseq_nice_add(kseq, nice);
}
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
}
Nice, is a property of a process as a whole.. I mistakenly moved it to the ksegroup when breaking up the process structure. Put it back in the proc structure.
2004-06-16 00:26:31 +00:00
}
p->p_nice = nice;
FOREACH_KSEGRP_IN_PROC(p, kg) {
sched_priority(kg);
FOREACH_THREAD_IN_GROUP(kg, td)
td->td_flags |= TDF_NEEDRESCHED;
}
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
}
void
Switch the sleep/wakeup and condition variable implementations to use the sleep queue interface: - Sleep queues attempt to merge some of the benefits of both sleep queues and condition variables. Having sleep qeueus in a hash table avoids having to allocate a queue head for each wait channel. Thus, struct cv has shrunk down to just a single char * pointer now. However, the hash table does not hold threads directly, but queue heads. This means that once you have located a queue in the hash bucket, you no longer have to walk the rest of the hash chain looking for threads. Instead, you have a list of all the threads sleeping on that wait channel. - Outside of the sleepq code and the sleep/cv code the kernel no longer differentiates between cv's and sleep/wakeup. For example, calls to abortsleep() and cv_abort() are replaced with a call to sleepq_abort(). Thus, the TDF_CVWAITQ flag is removed. Also, calls to unsleep() and cv_waitq_remove() have been replaced with calls to sleepq_remove(). - The sched_sleep() function no longer accepts a priority argument as sleep's no longer inherently bump the priority. Instead, this is soley a propery of msleep() which explicitly calls sched_prio() before blocking. - The TDF_ONSLEEPQ flag has been dropped as it was never used. The associated TDF_SET_ONSLEEPQ and TDF_CLR_ON_SLEEPQ macros have also been dropped and replaced with a single explicit clearing of td_wchan. TD_SET_ONSLEEPQ() would really have only made sense if it had taken the wait channel and message as arguments anyway. Now that that only happens in one place, a macro would be overkill.
2004-02-27 18:52:44 +00:00
sched_sleep(struct thread *td)
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
{
mtx_assert(&sched_lock, MA_OWNED);
td->td_slptime = ticks;
}
void
sched_wakeup(struct thread *td)
{
mtx_assert(&sched_lock, MA_OWNED);
/*
* Let the kseg know how long we slept for. This is because process
* interactivity behavior is modeled in the kseg.
*/
if (td->td_slptime) {
- Refrain from setting the td_priority in sched_wakeup(). It will be reset before we return to user space.
2003-03-03 04:11:40 +00:00
struct ksegrp *kg;
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
int hzticks;
- Refrain from setting the td_priority in sched_wakeup(). It will be reset before we return to user space.
2003-03-03 04:11:40 +00:00
kg = td->td_ksegrp;
- Change sched_interact_update() to only accept slp+runtime values between 0 and SCHED_SLP_RUN_MAX * 2. This allows us to simplify the algorithm quite a bit. Before, it dealt with arbitrary values which required us to do nasty integer division tricks that didn't quite work out correctly. - Chnage sched_wakeup() to detect conditions where the slp+runtime could exceed SCHED_SLP_RUN_MAX * 2. This can happen if we go to sleep for longer than 6 seconds. In this case, we'll just clear the runtime and set the sleep time to the max. - Define a new function, sched_interact_fork() which updates the slp+runtime of a newly forked thread. We want to limit the amount of history retained from the parent so that we learn the child's behavior quickly. We don't, however want to decay it to nothing. Previously, we would simply divide each parameter by 100 whenever we forked. After a few forks the values would reach 0 and tasks would not be considered interactive. - Add another KTR entry, cleanup some existing entries. - Remove a useless sched_interact_update() from sched_priority(). This is already done by the callers that require it.
2003-11-02 03:36:33 +00:00
hzticks = (ticks - td->td_slptime) << 10;
if (hzticks >= SCHED_SLP_RUN_MAX) {
kg->kg_slptime = SCHED_SLP_RUN_MAX;
kg->kg_runtime = 1;
} else {
kg->kg_slptime += hzticks;
sched_interact_update(kg);
}
- Refrain from setting the td_priority in sched_wakeup(). It will be reset before we return to user space.
2003-03-03 04:11:40 +00:00
sched_priority(kg);
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
sched_slice(td->td_kse);
- Refrain from setting the td_priority in sched_wakeup(). It will be reset before we return to user space.
2003-03-03 04:11:40 +00:00
td->td_slptime = 0;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
}
Give setrunqueue() and sched_add() more of a clue as to where they are coming from and what is expected from them. MFC after: 2 days
2004-09-01 02:11:28 +00:00
setrunqueue(td, SRQ_BORING);
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
}
/*
* Penalize the parent for creating a new child and initialize the child's
* priority.
*/
void
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
sched_fork(struct thread *td, struct thread *childtd)
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
{
mtx_assert(&sched_lock, MA_OWNED);
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
sched_fork_ksegrp(td, childtd->td_ksegrp);
sched_fork_thread(td, childtd);
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
}
void
When calling scheduler entrypoints for creating new threads and processes, specify "us" as the thread not the process/ksegrp/kse. You can always find the others from the thread but the converse is not true. Theorotically this would lead to runtime being allocated to the wrong entity in some cases though it is not clear how often this actually happenned. (would only affect threaded processes and would probably be pretty benign, but it WAS a bug..) Reviewed by: peter
2004-07-18 23:36:13 +00:00
sched_fork_ksegrp(struct thread *td, struct ksegrp *child)
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
{
When calling scheduler entrypoints for creating new threads and processes, specify "us" as the thread not the process/ksegrp/kse. You can always find the others from the thread but the converse is not true. Theorotically this would lead to runtime being allocated to the wrong entity in some cases though it is not clear how often this actually happenned. (would only affect threaded processes and would probably be pretty benign, but it WAS a bug..) Reviewed by: peter
2004-07-18 23:36:13 +00:00
struct ksegrp *kg = td->td_ksegrp;
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
mtx_assert(&sched_lock, MA_OWNED);
- Fix the maximum slice value. I accidentally checked in a value of '2' which meant no process would run for longer than 20ms. - Slightly redo the interactivity scorer. It follows the same algorithm but in a slightly more correct way. Previously values above half were incorrect. - Lower the interactivity threshold to 20. It seems that in testing non- interactive tasks are hardly ever near there and expensive interactive tasks can sometimes surpass it. This area needs more testing. - Remove an unnecessary KTR. - Fix a case where an idle thread that had an elevated priority due to priority prop. would be placed back on the idle queue. - Delay setting NEEDRESCHED until userret() for threads that haad their priority elevated while in kernel. This gives us the same context switch optimization as SCHED_4BSD. - Limit the child's slice to 1 in sched_fork_kse() so we detect its behavior more quickly. - Inhert some of the run/slp time from the child in sched_exit_ksegrp(). - Redo some of the priority comparisons so they are more clear. - Throttle the frequency of sched_pctcpu_update() so that rounding errors do not make it invalid.
2003-06-15 02:18:29 +00:00
- Change sched_interact_update() to only accept slp+runtime values between 0 and SCHED_SLP_RUN_MAX * 2. This allows us to simplify the algorithm quite a bit. Before, it dealt with arbitrary values which required us to do nasty integer division tricks that didn't quite work out correctly. - Chnage sched_wakeup() to detect conditions where the slp+runtime could exceed SCHED_SLP_RUN_MAX * 2. This can happen if we go to sleep for longer than 6 seconds. In this case, we'll just clear the runtime and set the sleep time to the max. - Define a new function, sched_interact_fork() which updates the slp+runtime of a newly forked thread. We want to limit the amount of history retained from the parent so that we learn the child's behavior quickly. We don't, however want to decay it to nothing. Previously, we would simply divide each parameter by 100 whenever we forked. After a few forks the values would reach 0 and tasks would not be considered interactive. - Add another KTR entry, cleanup some existing entries. - Remove a useless sched_interact_update() from sched_priority(). This is already done by the callers that require it.
2003-11-02 03:36:33 +00:00
child->kg_slptime = kg->kg_slptime;
child->kg_runtime = kg->kg_runtime;
child->kg_user_pri = kg->kg_user_pri;
sched_interact_fork(child);
- Add a new function "sched_interact_update()" that scales back the sleep and run time. - Scale the sleep and run time back via sched_interact_update() in more places. This is to keep the statistic more accurate. - Charge a parent one tick for forking a child. - Add only the run time and not the sleep time to the parents kg when a thread exits. This allows us to give a penalty for having an expensive thread exit but does not give a bonus for having an interactive thread exit. - Change the SLP_RUN_THROTTLE to limit us to 4/5th and not 1/2. - Change the SLP_RUN_MAX to two seconds. This keeps bursty interactive applications like mozilla and openoffice in the interactive range even through expensive tasks. - Recalculate the slice after every sleep. This ensures that once a task has been marked interactive it only has a slice of 1 at the risk of giving tasks that sleep for a very brief period a longer time slice.
2003-06-17 06:39:51 +00:00
kg->kg_runtime += tickincr << 10;
sched_interact_update(kg);
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
}
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
void
sched_fork_thread(struct thread *td, struct thread *child)
{
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
struct kse *ke;
struct kse *ke2;
sched_newthread(child);
ke = td->td_kse;
ke2 = child->td_kse;
ke2->ke_slice = 1; /* Attempt to quickly learn interactivity. */
ke2->ke_cpu = ke->ke_cpu;
ke2->ke_runq = NULL;
/* Grab our parents cpu estimation information. */
ke2->ke_ticks = ke->ke_ticks;
ke2->ke_ltick = ke->ke_ltick;
ke2->ke_ftick = ke->ke_ftick;
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
}
void
sched_class(struct ksegrp *kg, int class)
{
struct kseq *kseq;
struct kse *ke;
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
struct thread *td;
- Remove the ksq_loads[] array. We are only interested in three counts, the total load, the timeshare load, and the number of threads that can be migrated to another cpu. Account for these seperately. - Introduce a KSE_CAN_MIGRATE() macro which determines whether or not a KSE can be migrated to another CPU. Currently, this only checks to see if we're an interrupt handler. Eventually this will also be used to support CPU binding.
2003-11-02 10:56:48 +00:00
int nclass;
int oclass;
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
Add lock assertions for various proc/thread/kse/ksegroup fields to the scheduler functions.
2003-04-23 18:51:05 +00:00
mtx_assert(&sched_lock, MA_OWNED);
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
if (kg->kg_pri_class == class)
return;
- Remove the ksq_loads[] array. We are only interested in three counts, the total load, the timeshare load, and the number of threads that can be migrated to another cpu. Account for these seperately. - Introduce a KSE_CAN_MIGRATE() macro which determines whether or not a KSE can be migrated to another CPU. Currently, this only checks to see if we're an interrupt handler. Eventually this will also be used to support CPU binding.
2003-11-02 10:56:48 +00:00
nclass = PRI_BASE(class);
oclass = PRI_BASE(kg->kg_pri_class);
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
FOREACH_THREAD_IN_GROUP(kg, td) {
ke = td->td_kse;
- Add ke_runq == NULL to the conditions which will cause us to abort adjusting timeshare loads in sched_class(). This is only important if the thread has never run, otherwise the state checks should work as expected.
2005-02-04 17:22:46 +00:00
if ((ke->ke_state != KES_ONRUNQ &&
ke->ke_state != KES_THREAD) || ke->ke_runq == NULL)
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
continue;
kseq = KSEQ_CPU(ke->ke_cpu);
- Remove the ksq_loads[] array. We are only interested in three counts, the total load, the timeshare load, and the number of threads that can be migrated to another cpu. Account for these seperately. - Introduce a KSE_CAN_MIGRATE() macro which determines whether or not a KSE can be migrated to another CPU. Currently, this only checks to see if we're an interrupt handler. Eventually this will also be used to support CPU binding.
2003-11-02 10:56:48 +00:00
#ifdef SMP
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
/*
* On SMP if we're on the RUNQ we must adjust the transferable
* count because could be changing to or from an interrupt
* class.
*/
if (ke->ke_state == KES_ONRUNQ) {
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
if (KSE_CAN_MIGRATE(ke)) {
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
kseq->ksq_transferable--;
kseq->ksq_group->ksg_transferable--;
}
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
if (KSE_CAN_MIGRATE(ke)) {
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
kseq->ksq_transferable++;
kseq->ksq_group->ksg_transferable++;
}
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
}
- Remove the ksq_loads[] array. We are only interested in three counts, the total load, the timeshare load, and the number of threads that can be migrated to another cpu. Account for these seperately. - Introduce a KSE_CAN_MIGRATE() macro which determines whether or not a KSE can be migrated to another CPU. Currently, this only checks to see if we're an interrupt handler. Eventually this will also be used to support CPU binding.
2003-11-02 10:56:48 +00:00
#endif
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
if (oclass == PRI_TIMESHARE) {
- Remove the ksq_loads[] array. We are only interested in three counts, the total load, the timeshare load, and the number of threads that can be migrated to another cpu. Account for these seperately. - Introduce a KSE_CAN_MIGRATE() macro which determines whether or not a KSE can be migrated to another CPU. Currently, this only checks to see if we're an interrupt handler. Eventually this will also be used to support CPU binding.
2003-11-02 10:56:48 +00:00
kseq->ksq_load_timeshare--;
Nice, is a property of a process as a whole.. I mistakenly moved it to the ksegroup when breaking up the process structure. Put it back in the proc structure.
2004-06-16 00:26:31 +00:00
kseq_nice_rem(kseq, kg->kg_proc->p_nice);
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
}
if (nclass == PRI_TIMESHARE) {
kseq->ksq_load_timeshare++;
Nice, is a property of a process as a whole.. I mistakenly moved it to the ksegroup when breaking up the process structure. Put it back in the proc structure.
2004-06-16 00:26:31 +00:00
kseq_nice_add(kseq, kg->kg_proc->p_nice);
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
}
- Fix the ksq_load calculation. It now reflects the number of entries on the run queue for each cpu. - Introduce kse stealing into the sched_choose() code. This helps balance cpus better in cases where process turnover is high. This implementation is fairly trivial and will likely be only a temporary measure until something more sophisticated has been written.
2003-01-28 09:28:20 +00:00
}
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
kg->kg_pri_class = class;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
}
/*
* Return some of the child's priority and interactivity to the parent.
*/
void
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
sched_exit(struct proc *p, struct thread *childtd)
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
{
mtx_assert(&sched_lock, MA_OWNED);
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
sched_exit_ksegrp(FIRST_KSEGRP_IN_PROC(p), childtd);
- Remove earlier KTR_ULE tracepoints. - Define new KTR_SCHED points so that we can graph the operation of the scheduler.
2004-12-26 00:15:33 +00:00
sched_exit_thread(NULL, childtd);
- Add sched_exit_* - Call sched_exit_kse() from sched_exit() instead of implementing it here.
2003-04-11 19:24:00 +00:00
}
void
When calling scheduler entrypoints for creating new threads and processes, specify "us" as the thread not the process/ksegrp/kse. You can always find the others from the thread but the converse is not true. Theorotically this would lead to runtime being allocated to the wrong entity in some cases though it is not clear how often this actually happenned. (would only affect threaded processes and would probably be pretty benign, but it WAS a bug..) Reviewed by: peter
2004-07-18 23:36:13 +00:00
sched_exit_ksegrp(struct ksegrp *kg, struct thread *td)
- Add sched_exit_* - Call sched_exit_kse() from sched_exit() instead of implementing it here.
2003-04-11 19:24:00 +00:00
{
When calling scheduler entrypoints for creating new threads and processes, specify "us" as the thread not the process/ksegrp/kse. You can always find the others from the thread but the converse is not true. Theorotically this would lead to runtime being allocated to the wrong entity in some cases though it is not clear how often this actually happenned. (would only affect threaded processes and would probably be pretty benign, but it WAS a bug..) Reviewed by: peter
2004-07-18 23:36:13 +00:00
/* kg->kg_slptime += td->td_ksegrp->kg_slptime; */
kg->kg_runtime += td->td_ksegrp->kg_runtime;
- Add a new function "sched_interact_update()" that scales back the sleep and run time. - Scale the sleep and run time back via sched_interact_update() in more places. This is to keep the statistic more accurate. - Charge a parent one tick for forking a child. - Add only the run time and not the sleep time to the parents kg when a thread exits. This allows us to give a penalty for having an expensive thread exit but does not give a bonus for having an interactive thread exit. - Change the SLP_RUN_THROTTLE to limit us to 4/5th and not 1/2. - Change the SLP_RUN_MAX to two seconds. This keeps bursty interactive applications like mozilla and openoffice in the interactive range even through expensive tasks. - Recalculate the slice after every sleep. This ensures that once a task has been marked interactive it only has a slice of 1 at the risk of giving tasks that sleep for a very brief period a longer time slice.
2003-06-17 06:39:51 +00:00
sched_interact_update(kg);
- Add sched_exit_* - Call sched_exit_kse() from sched_exit() instead of implementing it here.
2003-04-11 19:24:00 +00:00
}
void
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
sched_exit_thread(struct thread *td, struct thread *childtd)
- Add sched_exit_* - Call sched_exit_kse() from sched_exit() instead of implementing it here.
2003-04-11 19:24:00 +00:00
{
- Remove earlier KTR_ULE tracepoints. - Define new KTR_SCHED points so that we can graph the operation of the scheduler.
2004-12-26 00:15:33 +00:00
CTR3(KTR_SCHED, "sched_exit_thread: %p(%s) prio %d",
childtd, childtd->td_proc->p_comm, childtd->td_priority);
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
kseq_load_rem(KSEQ_CPU(childtd->td_kse->ke_cpu), childtd->td_kse);
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
}
void
- Update the sched api. sched_{add,rem,clock,pctcpu} now all accept a td argument rather than a kse.
2003-10-16 08:39:15 +00:00
sched_clock(struct thread *td)
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
{
struct kseq *kseq;
- Use ksq_load as the authoritive count of kses on the pair of kseqs for sched_runnable() et all. - Remove some dead code in sched_clock(). - Define two macros KSEQ_SELF() and KSEQ_CPU() for getting the kseq of the current cpu or some alternate cpu. - Start introducing kseq_() functions, such as kseq_choose() and kseq_setup().
2003-01-29 07:00:51 +00:00
struct ksegrp *kg;
- Update the sched api. sched_{add,rem,clock,pctcpu} now all accept a td argument rather than a kse.
2003-10-16 08:39:15 +00:00
struct kse *ke;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
- Run sched_balance() and sched_balance_groups() from hardclock via sched_clock() rather than using callouts. This means we no longer have to take the load of the callout thread into consideration while balancing and should make the balancing decisions simpler and more accurate. Tested on: x86/UP, amd64/SMP
2004-06-02 05:46:48 +00:00
mtx_assert(&sched_lock, MA_OWNED);
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
kseq = KSEQ_SELF();
- Run sched_balance() and sched_balance_groups() from hardclock via sched_clock() rather than using callouts. This means we no longer have to take the load of the callout thread into consideration while balancing and should make the balancing decisions simpler and more accurate. Tested on: x86/UP, amd64/SMP
2004-06-02 05:46:48 +00:00
#ifdef SMP
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
if (ticks >= bal_tick)
- Run sched_balance() and sched_balance_groups() from hardclock via sched_clock() rather than using callouts. This means we no longer have to take the load of the callout thread into consideration while balancing and should make the balancing decisions simpler and more accurate. Tested on: x86/UP, amd64/SMP
2004-06-02 05:46:48 +00:00
sched_balance();
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
if (ticks >= gbal_tick && balance_groups)
- Run sched_balance() and sched_balance_groups() from hardclock via sched_clock() rather than using callouts. This means we no longer have to take the load of the callout thread into consideration while balancing and should make the balancing decisions simpler and more accurate. Tested on: x86/UP, amd64/SMP
2004-06-02 05:46:48 +00:00
sched_balance_groups();
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
/*
* We could have been assigned a non real-time thread without an
* IPI.
*/
if (kseq->ksq_assigned)
kseq_assign(kseq); /* Potentially sets NEEDRESCHED */
- Run sched_balance() and sched_balance_groups() from hardclock via sched_clock() rather than using callouts. This means we no longer have to take the load of the callout thread into consideration while balancing and should make the balancing decisions simpler and more accurate. Tested on: x86/UP, amd64/SMP
2004-06-02 05:46:48 +00:00
#endif
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
/*
* sched_setup() apparently happens prior to stathz being set. We
* need to resolve the timers earlier in the boot so we can avoid
* calculating this here.
*/
if (realstathz == 0) {
realstathz = stathz ? stathz : hz;
tickincr = hz / realstathz;
/*
* XXX This does not work for values of stathz that are much
* larger than hz.
*/
if (tickincr == 0)
tickincr = 1;
}
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
- Update the sched api. sched_{add,rem,clock,pctcpu} now all accept a td argument rather than a kse.
2003-10-16 08:39:15 +00:00
ke = td->td_kse;
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
kg = ke->ke_ksegrp;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
- Use ksq_load as the authoritive count of kses on the pair of kseqs for sched_runnable() et all. - Remove some dead code in sched_clock(). - Define two macros KSEQ_SELF() and KSEQ_CPU() for getting the kseq of the current cpu or some alternate cpu. - Start introducing kseq_() functions, such as kseq_choose() and kseq_setup().
2003-01-29 07:00:51 +00:00
/* Adjust ticks for pctcpu */
- Shift the tick count by 10 and back around sched_pctcpu_update() calculations. Keep this changes local to the function so the tick count is in its natural form otherwise. Previously 1000 was added each time a tick fired and we divided by 1000 when it was reported. This is done to reduce rounding errors.
2003-03-03 05:29:09 +00:00
ke->ke_ticks++;
- Allow idle's pctcpu time to be calculated.
2003-01-28 09:30:17 +00:00
ke->ke_ltick = ticks;
- Keep seperate statistics and run queues for different scheduling classes. - Treat each class specially in kseq_{choose,add,rem}. Let the rest of the code be less aware of scheduling classes. - Skip the interactivity calculation for non TIMESHARE ksegrps. - Move slice and runq selection into kseq_add(). Uninline it now that it's big.
2003-04-03 00:29:28 +00:00
- Allow idle's pctcpu time to be calculated.
2003-01-28 09:30:17 +00:00
/* Go up to one second beyond our max and then trim back down */
if (ke->ke_ftick + SCHED_CPU_TICKS + hz < ke->ke_ltick)
sched_pctcpu_update(ke);
Fix typo in last commit
2003-05-02 06:18:55 +00:00
if (td->td_flags & TDF_IDLETD)
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
return;
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
/*
- Use a better algorithm in sched_pctcpu_update() Contributed by: Thomaswuerfl@gmx.de - In sched_prio(), adjust the run queue for threads which may need to move to the current queue due to priority propagation . - In sched_switch(), fix style bug introduced when the KSE support went in. Columns are 80 chars wide, not 90. - In sched_switch(), Fix the comparison in the idle case and explicitly re-initialize the runq in the not propagated case. - Remove dead code in sched_clock(). - In sched_clock(), If we're an IDLE class td set NEEDRESCHED so that threads that have become runnable will get a chance to. - In sched_runnable(), if we're not the IDLETD, we should not consider curthread when examining the load. This mimics the 4BSD behavior of returning 0 when the only runnable thread is running. - In sched_userret(), remove the code for setting NEEDRESCHED entirely. This is not necessary and is not implemented in 4BSD. - Use the correct comparison in sched_add() when checking to see if an idle prio task has had it's priority temporarily elevated.
2003-10-27 06:47:05 +00:00
* We only do slicing code for TIMESHARE ksegrps.
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
*/
- Use a better algorithm in sched_pctcpu_update() Contributed by: Thomaswuerfl@gmx.de - In sched_prio(), adjust the run queue for threads which may need to move to the current queue due to priority propagation . - In sched_switch(), fix style bug introduced when the KSE support went in. Columns are 80 chars wide, not 90. - In sched_switch(), Fix the comparison in the idle case and explicitly re-initialize the runq in the not propagated case. - Remove dead code in sched_clock(). - In sched_clock(), If we're an IDLE class td set NEEDRESCHED so that threads that have become runnable will get a chance to. - In sched_runnable(), if we're not the IDLETD, we should not consider curthread when examining the load. This mimics the 4BSD behavior of returning 0 when the only runnable thread is running. - In sched_userret(), remove the code for setting NEEDRESCHED entirely. This is not necessary and is not implemented in 4BSD. - Use the correct comparison in sched_add() when checking to see if an idle prio task has had it's priority temporarily elevated.
2003-10-27 06:47:05 +00:00
if (kg->kg_pri_class != PRI_TIMESHARE)
return;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
/*
- Add a new variable 'kg_runtime' that tracks the amount of time we've run. - Use the ratio of kg_runtime / kg_slptime to determine our dynamic priority. - Scale kg_runtime and kg_slptime back when the sum of the two exceeds SCHED_SLP_RUN_MAX. This allows us to slowly forget old behavior. - Scale back the runtime and slptime in fork so that the new process has the same ratio but much less accumulated time. This causes new behavior to be noticed more quickly.
2003-02-10 14:03:45 +00:00
* We used a tick charge it to the ksegrp so that we can compute our
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
* interactivity.
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
*/
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
kg->kg_runtime += tickincr << 10;
- Add a new function "sched_interact_update()" that scales back the sleep and run time. - Scale the sleep and run time back via sched_interact_update() in more places. This is to keep the statistic more accurate. - Charge a parent one tick for forking a child. - Add only the run time and not the sleep time to the parents kg when a thread exits. This allows us to give a penalty for having an expensive thread exit but does not give a bonus for having an interactive thread exit. - Change the SLP_RUN_THROTTLE to limit us to 4/5th and not 1/2. - Change the SLP_RUN_MAX to two seconds. This keeps bursty interactive applications like mozilla and openoffice in the interactive range even through expensive tasks. - Recalculate the slice after every sleep. This ensures that once a task has been marked interactive it only has a slice of 1 at the risk of giving tasks that sleep for a very brief period a longer time slice.
2003-06-17 06:39:51 +00:00
sched_interact_update(kg);
- Add a new variable 'kg_runtime' that tracks the amount of time we've run. - Use the ratio of kg_runtime / kg_slptime to determine our dynamic priority. - Scale kg_runtime and kg_slptime back when the sum of the two exceeds SCHED_SLP_RUN_MAX. This allows us to slowly forget old behavior. - Scale back the runtime and slptime in fork so that the new process has the same ratio but much less accumulated time. This causes new behavior to be noticed more quickly.
2003-02-10 14:03:45 +00:00
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
/*
* We used up one time slice.
*/
- Remove long dead code. rslices hasn't been used in some time and neither has sched_pickcpu().
2003-11-17 08:24:14 +00:00
if (--ke->ke_slice > 0)
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
return;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
/*
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
* We're out of time, recompute priorities and requeue.
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
*/
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
kseq_load_rem(kseq, ke);
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
sched_priority(kg);
sched_slice(ke);
if (SCHED_CURR(kg, ke))
ke->ke_runq = kseq->ksq_curr;
else
ke->ke_runq = kseq->ksq_next;
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
kseq_load_add(kseq, ke);
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
td->td_flags |= TDF_NEEDRESCHED;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
}
int
sched_runnable(void)
{
struct kseq *kseq;
- When a new thread is added to a kseq the load is incremented prior to adding it to the nice tables. Therefore, in kseq_add_nice, we should keep in mind that the load will be 1 if we are the only thread, and not 0. - Assert that the sched lock is held in all the appropriate places. - Increase the scope of the sched lock in sched_pctcpu_update(). - Hold the sched lock in sched_runnable(). It is not held by the caller.
2003-06-08 00:47:33 +00:00
int load;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
- When a new thread is added to a kseq the load is incremented prior to adding it to the nice tables. Therefore, in kseq_add_nice, we should keep in mind that the load will be 1 if we are the only thread, and not 0. - Assert that the sched lock is held in all the appropriate places. - Increase the scope of the sched lock in sched_pctcpu_update(). - Hold the sched lock in sched_runnable(). It is not held by the caller.
2003-06-08 00:47:33 +00:00
load = 1;
- Use ksq_load as the authoritive count of kses on the pair of kseqs for sched_runnable() et all. - Remove some dead code in sched_clock(). - Define two macros KSEQ_SELF() and KSEQ_CPU() for getting the kseq of the current cpu or some alternate cpu. - Start introducing kseq_() functions, such as kseq_choose() and kseq_setup().
2003-01-29 07:00:51 +00:00
kseq = KSEQ_SELF();
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
#ifdef SMP
- It's ok if sched_runnable() has races in it, we don't need the sched_lock here unless we have something on the assigned queue.
2003-11-05 05:30:12 +00:00
if (kseq->ksq_assigned) {
mtx_lock_spin(&sched_lock);
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
kseq_assign(kseq);
- It's ok if sched_runnable() has races in it, we don't need the sched_lock here unless we have something on the assigned queue.
2003-11-05 05:30:12 +00:00
mtx_unlock_spin(&sched_lock);
}
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
#endif
- Use a better algorithm in sched_pctcpu_update() Contributed by: Thomaswuerfl@gmx.de - In sched_prio(), adjust the run queue for threads which may need to move to the current queue due to priority propagation . - In sched_switch(), fix style bug introduced when the KSE support went in. Columns are 80 chars wide, not 90. - In sched_switch(), Fix the comparison in the idle case and explicitly re-initialize the runq in the not propagated case. - Remove dead code in sched_clock(). - In sched_clock(), If we're an IDLE class td set NEEDRESCHED so that threads that have become runnable will get a chance to. - In sched_runnable(), if we're not the IDLETD, we should not consider curthread when examining the load. This mimics the 4BSD behavior of returning 0 when the only runnable thread is running. - In sched_userret(), remove the code for setting NEEDRESCHED entirely. This is not necessary and is not implemented in 4BSD. - Use the correct comparison in sched_add() when checking to see if an idle prio task has had it's priority temporarily elevated.
2003-10-27 06:47:05 +00:00
if ((curthread->td_flags & TDF_IDLETD) != 0) {
if (kseq->ksq_load > 0)
goto out;
} else
if (kseq->ksq_load - 1 > 0)
goto out;
- When a new thread is added to a kseq the load is incremented prior to adding it to the nice tables. Therefore, in kseq_add_nice, we should keep in mind that the load will be 1 if we are the only thread, and not 0. - Assert that the sched lock is held in all the appropriate places. - Increase the scope of the sched lock in sched_pctcpu_update(). - Hold the sched lock in sched_runnable(). It is not held by the caller.
2003-06-08 00:47:33 +00:00
load = 0;
out:
return (load);
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
}
void
sched_userret(struct thread *td)
{
struct ksegrp *kg;
Rework the interface between priority propagation (lending) and the schedulers a bit to ensure more correct handling of priorities and fewer priority inversions: - Add two functions to the sched(9) API to handle priority lending: sched_lend_prio() and sched_unlend_prio(). The turnstile code uses these functions to ask the scheduler to lend a thread a set priority and to tell the scheduler when it thinks it is ok for a thread to stop borrowing priority. The unlend case is slightly complex in that the turnstile code tells the scheduler what the minimum priority of the thread needs to be to satisfy the requirements of any other threads blocked on locks owned by the thread in question. The scheduler then decides where the thread can go back to normal mode (if it's normal priority is high enough to satisfy the pending lock requests) or it it should continue to use the priority specified to the sched_unlend_prio() call. This involves adding a new per-thread flag TDF_BORROWING that replaces the ULE-only kse flag for priority elevation. - Schedulers now refuse to lower the priority of a thread that is currently borrowing another therad's priority. - If a scheduler changes the priority of a thread that is currently sitting on a turnstile, it will call a new function turnstile_adjust() to inform the turnstile code of the change. This function resorts the thread on the priority list of the turnstile if needed, and if the thread ends up at the head of the list (due to having the highest priority) and its priority was raised, then it will propagate that new priority to the owner of the lock it is blocked on. Some additional fixes specific to the 4BSD scheduler include: - Common code for updating the priority of a thread when the user priority of its associated kse group has been consolidated in a new static function resetpriority_thread(). One change to this function is that it will now only adjust the priority of a thread if it already has a time sharing priority, thus preserving any boosts from a tsleep() until the thread returns to userland. Also, resetpriority() no longer calls maybe_resched() on each thread in the group. Instead, the code calling resetpriority() is responsible for calling resetpriority_thread() on any threads that need to be updated. - schedcpu() now uses resetpriority_thread() instead of just calling sched_prio() directly after it updates a kse group's user priority. - sched_clock() now uses resetpriority_thread() rather than writing directly to td_priority. - sched_nice() now updates all the priorities of the threads after the group priority has been adjusted. Discussed with: bde Reviewed by: ups, jeffr Tested on: 4bsd, ule Tested on: i386, alpha, sparc64
2004-12-30 20:52:44 +00:00
KASSERT((td->td_flags & TDF_BORROWING) == 0,
("thread with borrowed priority returning to userland"));
kg = td->td_ksegrp;
if (td->td_priority != kg->kg_user_pri) {
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
mtx_lock_spin(&sched_lock);
td->td_priority = kg->kg_user_pri;
Rework the interface between priority propagation (lending) and the schedulers a bit to ensure more correct handling of priorities and fewer priority inversions: - Add two functions to the sched(9) API to handle priority lending: sched_lend_prio() and sched_unlend_prio(). The turnstile code uses these functions to ask the scheduler to lend a thread a set priority and to tell the scheduler when it thinks it is ok for a thread to stop borrowing priority. The unlend case is slightly complex in that the turnstile code tells the scheduler what the minimum priority of the thread needs to be to satisfy the requirements of any other threads blocked on locks owned by the thread in question. The scheduler then decides where the thread can go back to normal mode (if it's normal priority is high enough to satisfy the pending lock requests) or it it should continue to use the priority specified to the sched_unlend_prio() call. This involves adding a new per-thread flag TDF_BORROWING that replaces the ULE-only kse flag for priority elevation. - Schedulers now refuse to lower the priority of a thread that is currently borrowing another therad's priority. - If a scheduler changes the priority of a thread that is currently sitting on a turnstile, it will call a new function turnstile_adjust() to inform the turnstile code of the change. This function resorts the thread on the priority list of the turnstile if needed, and if the thread ends up at the head of the list (due to having the highest priority) and its priority was raised, then it will propagate that new priority to the owner of the lock it is blocked on. Some additional fixes specific to the 4BSD scheduler include: - Common code for updating the priority of a thread when the user priority of its associated kse group has been consolidated in a new static function resetpriority_thread(). One change to this function is that it will now only adjust the priority of a thread if it already has a time sharing priority, thus preserving any boosts from a tsleep() until the thread returns to userland. Also, resetpriority() no longer calls maybe_resched() on each thread in the group. Instead, the code calling resetpriority() is responsible for calling resetpriority_thread() on any threads that need to be updated. - schedcpu() now uses resetpriority_thread() instead of just calling sched_prio() directly after it updates a kse group's user priority. - sched_clock() now uses resetpriority_thread() rather than writing directly to td_priority. - sched_nice() now updates all the priorities of the threads after the group priority has been adjusted. Discussed with: bde Reviewed by: ups, jeffr Tested on: 4bsd, ule Tested on: i386, alpha, sparc64
2004-12-30 20:52:44 +00:00
td->td_base_pri = kg->kg_user_pri;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
mtx_unlock_spin(&sched_lock);
}
}
- Fix the ksq_load calculation. It now reflects the number of entries on the run queue for each cpu. - Introduce kse stealing into the sched_choose() code. This helps balance cpus better in cases where process turnover is high. This implementation is fairly trivial and will likely be only a temporary measure until something more sophisticated has been written.
2003-01-28 09:28:20 +00:00
struct kse *
sched_choose(void)
{
- Use ksq_load as the authoritive count of kses on the pair of kseqs for sched_runnable() et all. - Remove some dead code in sched_clock(). - Define two macros KSEQ_SELF() and KSEQ_CPU() for getting the kseq of the current cpu or some alternate cpu. - Start introducing kseq_() functions, such as kseq_choose() and kseq_setup().
2003-01-29 07:00:51 +00:00
struct kseq *kseq;
- Fix the ksq_load calculation. It now reflects the number of entries on the run queue for each cpu. - Introduce kse stealing into the sched_choose() code. This helps balance cpus better in cases where process turnover is high. This implementation is fairly trivial and will likely be only a temporary measure until something more sophisticated has been written.
2003-01-28 09:28:20 +00:00
struct kse *ke;
- When a new thread is added to a kseq the load is incremented prior to adding it to the nice tables. Therefore, in kseq_add_nice, we should keep in mind that the load will be 1 if we are the only thread, and not 0. - Assert that the sched lock is held in all the appropriate places. - Increase the scope of the sched lock in sched_pctcpu_update(). - Hold the sched lock in sched_runnable(). It is not held by the caller.
2003-06-08 00:47:33 +00:00
mtx_assert(&sched_lock, MA_OWNED);
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
kseq = KSEQ_SELF();
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
#ifdef SMP
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
restart:
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
if (kseq->ksq_assigned)
kseq_assign(kseq);
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
#endif
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
ke = kseq_choose(kseq);
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
if (ke) {
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
#ifdef SMP
if (ke->ke_ksegrp->kg_pri_class == PRI_IDLE)
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
if (kseq_idled(kseq) == 0)
goto restart;
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
#endif
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
kseq_runq_rem(kseq, ke);
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
ke->ke_state = KES_THREAD;
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
return (ke);
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
}
- Fix the ksq_load calculation. It now reflects the number of entries on the run queue for each cpu. - Introduce kse stealing into the sched_choose() code. This helps balance cpus better in cases where process turnover is high. This implementation is fairly trivial and will likely be only a temporary measure until something more sophisticated has been written.
2003-01-28 09:28:20 +00:00
#ifdef SMP
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
if (kseq_idled(kseq) == 0)
goto restart;
- Fix the ksq_load calculation. It now reflects the number of entries on the run queue for each cpu. - Introduce kse stealing into the sched_choose() code. This helps balance cpus better in cases where process turnover is high. This implementation is fairly trivial and will likely be only a temporary measure until something more sophisticated has been written.
2003-01-28 09:28:20 +00:00
#endif
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
return (NULL);
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
}
void
Give setrunqueue() and sched_add() more of a clue as to where they are coming from and what is expected from them. MFC after: 2 days
2004-09-01 02:11:28 +00:00
sched_add(struct thread *td, int flags)
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
{
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
struct kseq *kseq;
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
struct ksegrp *kg;
- Update the sched api. sched_{add,rem,clock,pctcpu} now all accept a td argument rather than a kse.
2003-10-16 08:39:15 +00:00
struct kse *ke;
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
int preemptive;
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
int canmigrate;
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
int class;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
- Remove earlier KTR_ULE tracepoints. - Define new KTR_SCHED points so that we can graph the operation of the scheduler.
2004-12-26 00:15:33 +00:00
CTR5(KTR_SCHED, "sched_add: %p(%s) prio %d by %p(%s)",
td, td->td_proc->p_comm, td->td_priority, curthread,
curthread->td_proc->p_comm);
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
mtx_assert(&sched_lock, MA_OWNED);
- Update the sched api. sched_{add,rem,clock,pctcpu} now all accept a td argument rather than a kse.
2003-10-16 08:39:15 +00:00
ke = td->td_kse;
kg = td->td_ksegrp;
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
canmigrate = 1;
preemptive = !(flags & SRQ_YIELDING);
class = PRI_BASE(kg->kg_pri_class);
kseq = KSEQ_SELF();
if ((ke->ke_flags & KEF_INTERNAL) == 0)
SLOT_USE(td->td_ksegrp);
ke->ke_flags &= ~KEF_INTERNAL;
#ifdef SMP
- Take up a 'slot' while we're on the assigned queue, waiting to be posted to another processor. Otherwise, kern_switch() gets confused and tries to sched_add(NULL).
2004-12-13 13:09:33 +00:00
if (ke->ke_flags & KEF_ASSIGNED) {
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
if (ke->ke_flags & KEF_REMOVED)
- Take up a 'slot' while we're on the assigned queue, waiting to be posted to another processor. Otherwise, kern_switch() gets confused and tries to sched_add(NULL).
2004-12-13 13:09:33 +00:00
ke->ke_flags &= ~KEF_REMOVED;
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
return;
- Take up a 'slot' while we're on the assigned queue, waiting to be posted to another processor. Otherwise, kern_switch() gets confused and tries to sched_add(NULL).
2004-12-13 13:09:33 +00:00
}
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
canmigrate = KSE_CAN_MIGRATE(ke);
#endif
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
KASSERT(ke->ke_state != KES_ONRUNQ,
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
("sched_add: kse %p (%s) already in run queue", ke,
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
ke->ke_proc->p_comm));
KASSERT(ke->ke_proc->p_sflag & PS_INMEM,
- Make some context switches conditional on SCHED_STRICT_RESCHED. This may have some negative effect on interactivity but it yields great perf. gains. This also brings the conditions under which ULE context switches inline with SCHED_4BSD. - Define some new kseq_* functions for manipulating the run queue. - Add a new kseq member ksq_rslices and ksq_bload. rslices is the sum of the slices of runnable kses. This will be used for push load balance decisions. bload is the number of threads blocked waiting on IO.
2003-02-03 05:30:07 +00:00
("sched_add: process swapped out"));
- Clean up some debug code left over from my earlier megacommit.
2003-04-12 07:28:36 +00:00
KASSERT(ke->ke_runq == NULL,
("sched_add: KSE %p is still assigned to a run queue", ke));
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
switch (class) {
- Keep seperate statistics and run queues for different scheduling classes. - Treat each class specially in kseq_{choose,add,rem}. Let the rest of the code be less aware of scheduling classes. - Skip the interactivity calculation for non TIMESHARE ksegrps. - Move slice and runq selection into kseq_add(). Uninline it now that it's big.
2003-04-03 00:29:28 +00:00
case PRI_ITHD:
case PRI_REALTIME:
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
ke->ke_runq = kseq->ksq_curr;
ke->ke_slice = SCHED_SLICE_MAX;
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
if (canmigrate)
ke->ke_cpu = PCPU_GET(cpuid);
- Keep seperate statistics and run queues for different scheduling classes. - Treat each class specially in kseq_{choose,add,rem}. Let the rest of the code be less aware of scheduling classes. - Skip the interactivity calculation for non TIMESHARE ksegrps. - Move slice and runq selection into kseq_add(). Uninline it now that it's big.
2003-04-03 00:29:28 +00:00
break;
case PRI_TIMESHARE:
- Clean up some debug code left over from my earlier megacommit.
2003-04-12 07:28:36 +00:00
if (SCHED_CURR(kg, ke))
ke->ke_runq = kseq->ksq_curr;
else
ke->ke_runq = kseq->ksq_next;
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
break;
- Keep seperate statistics and run queues for different scheduling classes. - Treat each class specially in kseq_{choose,add,rem}. Let the rest of the code be less aware of scheduling classes. - Skip the interactivity calculation for non TIMESHARE ksegrps. - Move slice and runq selection into kseq_add(). Uninline it now that it's big.
2003-04-03 00:29:28 +00:00
case PRI_IDLE:
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
/*
* This is for priority prop.
*/
- Use a better algorithm in sched_pctcpu_update() Contributed by: Thomaswuerfl@gmx.de - In sched_prio(), adjust the run queue for threads which may need to move to the current queue due to priority propagation . - In sched_switch(), fix style bug introduced when the KSE support went in. Columns are 80 chars wide, not 90. - In sched_switch(), Fix the comparison in the idle case and explicitly re-initialize the runq in the not propagated case. - Remove dead code in sched_clock(). - In sched_clock(), If we're an IDLE class td set NEEDRESCHED so that threads that have become runnable will get a chance to. - In sched_runnable(), if we're not the IDLETD, we should not consider curthread when examining the load. This mimics the 4BSD behavior of returning 0 when the only runnable thread is running. - In sched_userret(), remove the code for setting NEEDRESCHED entirely. This is not necessary and is not implemented in 4BSD. - Use the correct comparison in sched_add() when checking to see if an idle prio task has had it's priority temporarily elevated.
2003-10-27 06:47:05 +00:00
if (ke->ke_thread->td_priority < PRI_MIN_IDLE)
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
ke->ke_runq = kseq->ksq_curr;
else
ke->ke_runq = &kseq->ksq_idle;
ke->ke_slice = SCHED_SLICE_MIN;
break;
default:
- Change sched_interact_update() to only accept slp+runtime values between 0 and SCHED_SLP_RUN_MAX * 2. This allows us to simplify the algorithm quite a bit. Before, it dealt with arbitrary values which required us to do nasty integer division tricks that didn't quite work out correctly. - Chnage sched_wakeup() to detect conditions where the slp+runtime could exceed SCHED_SLP_RUN_MAX * 2. This can happen if we go to sleep for longer than 6 seconds. In this case, we'll just clear the runtime and set the sleep time to the max. - Define a new function, sched_interact_fork() which updates the slp+runtime of a newly forked thread. We want to limit the amount of history retained from the parent so that we learn the child's behavior quickly. We don't, however want to decay it to nothing. Previously, we would simply divide each parameter by 100 whenever we forked. After a few forks the values would reach 0 and tasks would not be considered interactive. - Add another KTR entry, cleanup some existing entries. - Remove a useless sched_interact_update() from sched_priority(). This is already done by the callers that require it.
2003-11-02 03:36:33 +00:00
panic("Unknown pri class.");
- Keep seperate statistics and run queues for different scheduling classes. - Treat each class specially in kseq_{choose,add,rem}. Let the rest of the code be less aware of scheduling classes. - Skip the interactivity calculation for non TIMESHARE ksegrps. - Move slice and runq selection into kseq_add(). Uninline it now that it's big.
2003-04-03 00:29:28 +00:00
break;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
}
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
#ifdef SMP
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
/*
* Don't migrate running threads here. Force the long term balancer
* to do it.
*/
- Introduce a new flag KEF_HOLD that prevents sched_add() from doing a migration. Use this in sched_prio() and sched_switch() to stop us from migrating threads that are in short term sleeps or are runnable. These extra migrations were added in the patches to support KSE. - Only set NEEDRESCHED if the thread we're adding in sched_add() is a lower priority and is being placed on the current queue. - Fix some minor whitespace problems.
2004-08-12 07:56:33 +00:00
if (ke->ke_flags & KEF_HOLD) {
ke->ke_flags &= ~KEF_HOLD;
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
canmigrate = 0;
- Introduce a new flag KEF_HOLD that prevents sched_add() from doing a migration. Use this in sched_prio() and sched_switch() to stop us from migrating threads that are in short term sleeps or are runnable. These extra migrations were added in the patches to support KSE. - Only set NEEDRESCHED if the thread we're adding in sched_add() is a lower priority and is being placed on the current queue. - Fix some minor whitespace problems.
2004-08-12 07:56:33 +00:00
}
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
/*
* If this thread is pinned or bound, notify the target cpu.
*/
if (!canmigrate && ke->ke_cpu != PCPU_GET(cpuid) ) {
- Assign the ke_cpu field in kseq_notify() so that all of our callers do not have to do it. - Set the ke_runq to NULL in sched_add() before calling kseq_notify(). Otherwise we may panic in sched_add() if INVARIANTS is on.
2003-12-14 02:06:29 +00:00
ke->ke_runq = NULL;
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
kseq_notify(ke, ke->ke_cpu);
return;
}
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
/*
- In kseq_transfer() return if smp has not been started. - In sched_add(), do the idle check prior to the transfer check so that we don't try to transfer load from an idle cpu. This fixes panics caused by IPIs on UP machines running SMP kernels. Reported/Debugged by: seanc
2003-12-20 14:03:14 +00:00
* If we had been idle, clear our bit in the group and potentially
* the global bitmap. If not, see if we should transfer this thread.
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
*/
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
if ((class == PRI_TIMESHARE || class == PRI_REALTIME) &&
(kseq->ksq_group->ksg_idlemask & PCPU_GET(cpumask)) != 0) {
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
/*
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
* Check to see if our group is unidling, and if so, remove it
* from the global idle mask.
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
*/
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
if (kseq->ksq_group->ksg_idlemask ==
kseq->ksq_group->ksg_cpumask)
atomic_clear_int(&kseq_idle, kseq->ksq_group->ksg_mask);
/*
* Now remove ourselves from the group specific idle mask.
*/
kseq->ksq_group->ksg_idlemask &= ~PCPU_GET(cpumask);
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
} else if (canmigrate && kseq->ksq_load > 1 && class != PRI_ITHD)
- In kseq_transfer() return if smp has not been started. - In sched_add(), do the idle check prior to the transfer check so that we don't try to transfer load from an idle cpu. This fixes panics caused by IPIs on UP machines running SMP kernels. Reported/Debugged by: seanc
2003-12-20 14:03:14 +00:00
if (kseq_transfer(kseq, ke, class))
return;
- Use a new flag, KEF_XFERABLE, to record with certainty that this kse had contributed to the transferable load count. This prevents any potential problems with sched_pin() being used around calls to setrunqueue(). - Change the sched_add() load balancing algorithm to try to migrate on wakeup. This attempts to place threads that communicate with each other on the same CPU. - Don't clear the idle counts in kseq_transfer(), let the cpus do that when they call sched_add() from kseq_assign(). - Correct a few out of date comments. - Make sure the ke_cpu field is correct when we preempt. - Call kseq_assign() from sched_clock() to catch any assignments that were done without IPI. Presently all assignments are done with an IPI, but I'm trying a patch that limits that. - Don't migrate a thread if it is still runnable in sched_add(). Previously, this could only happen for KSE threads, but due to changes to sched_switch() all threads went through this path. - Remove some code that was added with preemption but is not necessary.
2004-08-10 07:52:21 +00:00
ke->ke_cpu = PCPU_GET(cpuid);
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
#endif
- Introduce a new flag KEF_HOLD that prevents sched_add() from doing a migration. Use this in sched_prio() and sched_switch() to stop us from migrating threads that are in short term sleeps or are runnable. These extra migrations were added in the patches to support KSE. - Only set NEEDRESCHED if the thread we're adding in sched_add() is a lower priority and is being placed on the current queue. - Fix some minor whitespace problems.
2004-08-12 07:56:33 +00:00
if (td->td_priority < curthread->td_priority &&
ke->ke_runq == kseq->ksq_curr)
curthread->td_flags |= TDF_NEEDRESCHED;
- Move contents of sched_add() into a sched_add_internal() function that takes an argument to specify if it should preempt or not. Don't preempt when sched_add_internal() is called from kseq_idled() or kseq_assign() as in those cases we are about to call mi_switch() anyways. Also, doing so during the first context switch on an AP leads to a NULL pointer deref because curthread is NULL. - Reenable preemption for ULE. Submitted by: Taku YAMAMOTO taku at tackymt.homeip.net
2004-07-08 21:45:04 +00:00
if (preemptive && maybe_preempt(td))
Implement preemption of kernel threads natively in the scheduler rather than as one-off hacks in various other parts of the kernel: - Add a function maybe_preempt() that is called from sched_add() to determine if a thread about to be added to a run queue should be preempted to directly. If it is not safe to preempt or if the new thread does not have a high enough priority, then the function returns false and sched_add() adds the thread to the run queue. If the thread should be preempted to but the current thread is in a nested critical section, then the flag TDF_OWEPREEMPT is set and the thread is added to the run queue. Otherwise, mi_switch() is called immediately and the thread is never added to the run queue since it is switch to directly. When exiting an outermost critical section, if TDF_OWEPREEMPT is set, then clear it and call mi_switch() to perform the deferred preemption. - Remove explicit preemption from ithread_schedule() as calling setrunqueue() now does all the correct work. This also removes the do_switch argument from ithread_schedule(). - Do not use the manual preemption code in mtx_unlock if the architecture supports native preemption. - Don't call mi_switch() in a loop during shutdown to give ithreads a chance to run if the architecture supports native preemption since the ithreads will just preempt DELAY(). - Don't call mi_switch() from the page zeroing idle thread for architectures that support native preemption as it is unnecessary. - Native preemption is enabled on the same archs that supported ithread preemption, namely alpha, i386, and amd64. This change should largely be a NOP for the default case as committed except that we will do fewer context switches in a few cases and will avoid the run queues completely when preempting. Approved by: scottl (with his re@ hat)
2004-07-02 20:21:44 +00:00
return;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
ke->ke_state = KES_ONRUNQ;
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
kseq_runq_add(kseq, ke, flags);
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
kseq_load_add(kseq, ke);
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
}
void
- Update the sched api. sched_{add,rem,clock,pctcpu} now all accept a td argument rather than a kse.
2003-10-16 08:39:15 +00:00
sched_rem(struct thread *td)
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
{
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
struct kseq *kseq;
- Update the sched api. sched_{add,rem,clock,pctcpu} now all accept a td argument rather than a kse.
2003-10-16 08:39:15 +00:00
struct kse *ke;
- Remove earlier KTR_ULE tracepoints. - Define new KTR_SCHED points so that we can graph the operation of the scheduler.
2004-12-26 00:15:33 +00:00
CTR5(KTR_SCHED, "sched_rem: %p(%s) prio %d by %p(%s)",
td, td->td_proc->p_comm, td->td_priority, curthread,
curthread->td_proc->p_comm);
- Fix a long standing problem where an ithread would not honor sched_pin(). - Remove the sched_add wrapper that used sched_add_internal() as a backend. Its only purpose was to interpret one flag and turn it into an int. Do the right thing and interpret the flag in sched_add() instead. - Pass the flag argument to sched_add() to kseq_runq_add() so that we can get the SRQ_PREEMPT optimization too. - Add a KEF_INTERNAL flag. If KEF_INTERNAL is set we don't adjust the SLOT counts, otherwise the slot counts are adjusted as soon as we enter sched_add() or sched_rem() rather than when the thread is actually placed on the run queue. This greatly simplifies the handling of slots. - Remove the explicit prevention of migration for ithreads on non-x86 platforms. This was never shown to have any real benefit. - Remove the unused class argument to KSE_CAN_MIGRATE(). - Add ktr points for thread migration events. - Fix a long standing bug on platforms which don't initialize the cpu topology. The ksg_maxid variable was never correctly set on these platforms which caused the long term load balancer to never inspect more than the first group or processor. - Fix another bug which prevented the long term load balancer from working properly. If stathz != hz we can't expect sched_clock() to be called on the exact tick count that we're anticipating. - Rearrange sched_switch() a bit to reduce indentation levels.
2004-12-26 22:56:08 +00:00
mtx_assert(&sched_lock, MA_OWNED);
ke = td->td_kse;
SLOT_RELEASE(td->td_ksegrp);
- Take up a 'slot' while we're on the assigned queue, waiting to be posted to another processor. Otherwise, kern_switch() gets confused and tries to sched_add(NULL).
2004-12-13 13:09:33 +00:00
if (ke->ke_flags & KEF_ASSIGNED) {
ke->ke_flags |= KEF_REMOVED;
- Add static to local functions and data where it was missing. - Add an IPI based mechanism for migrating kses. This mechanism is broken down into several components. This is intended to reduce cache thrashing by eliminating most cases where one cpu touches another's run queues. - kseq_notify() appends a kse to a lockless singly linked list and conditionally sends an IPI to the target processor. Right now this is protected by sched_lock but at some point I'd like to get rid of the global lock. This is why I used something more complicated than a standard queue. - kseq_assign() processes our list of kses that have been assigned to us by other processors. This simply calls sched_add() for each item on the list after clearing the new KEF_ASSIGNED flag. This flag is used to indicate that we have been appeneded to the assigned queue but not added to the run queue yet. - In sched_add(), instead of adding a KSE to another processor's queue we use kse_notify() so that we don't touch their queue. Also in sched_add(), if KEF_ASSIGNED is already set return immediately. This can happen if a thread is removed and readded so that the priority is recorded properly. - In sched_rem() return immediately if KEF_ASSIGNED is set. All callers immediately readd simply to adjust priorites etc. - In sched_choose(), if we're running an IDLE task or the per cpu idle thread set our cpumask bit in 'kseq_idle' so that other processors may know that we are idle. Before this, make a single pass through the run queues of other processors so that we may find work more immediately if it is available. - In sched_runnable(), don't scan each processor's run queue, they will IPI us if they have work for us to do. - In sched_add(), if we're adding a thread that can be migrated and we have plenty of work to do, try to migrate the thread to an idle kseq. - Simplify the logic in sched_prio() and take the KEF_ASSIGNED flag into consideration. - No longer use kseq_choose() to steal threads, it can lose it's last argument. - Create a new function runq_steal() which operates like runq_choose() but skips threads based on some criteria. Currently it will not steal PRI_ITHD threads. In the future this will be used for CPU binding. - Create a kseq_steal() that checks each run queue with runq_steal(), use kseq_steal() in the places where we used kseq_choose() to steal with before.
2003-10-31 11:16:04 +00:00
return;
- Take up a 'slot' while we're on the assigned queue, waiting to be posted to another processor. Otherwise, kern_switch() gets confused and tries to sched_add(NULL).
2004-12-13 13:09:33 +00:00
}
- Clean up KASSERTS.
2004-01-25 08:57:38 +00:00
KASSERT((ke->ke_state == KES_ONRUNQ),
("sched_rem: KSE not on run queue"));
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
- Take up a 'slot' while we're on the assigned queue, waiting to be posted to another processor. Otherwise, kern_switch() gets confused and tries to sched_add(NULL).
2004-12-13 13:09:33 +00:00
ke->ke_state = KES_THREAD;
- Add a SYSCTL node for the ule scheduler. - Allow user adjustable min and max time slices (suggested by hiten). - Change the SLP_RUN_MAX to 100ms from 2 seconds so that we learn whether a process is interactive or not much more quickly. - Place a process on the current run queue if it is interactive or if it is running at an interrupt thread priority due to priority prop. - Use the 'current' timeshare queue for interrupt threads, realtime threads, and idle threads that are running at higher priority due to priority prop. This fixes problems where priorities would have been elevated but we would not check the timeshare run queue until other lower priority tasks were no longer runnable. - Keep an array of loads indexed by the priority class as well as a global load. - Keep an bucket of nice values with a count of the number of kses currently runnable with that nice value. - Keep track of the minimum nice value of any running thread. - Remove the unused short term sleep accounting. I was attempting to use this for load balancing but it didn't work out. - Define a kseq_print() for use with debugging. - Add KTR debugging at useful places so we can easily debug slice and priority assignment. - Decouple the runq assignment from the kseq assignment. kseq_add now keeps track of statistics. This is done so that the nice and load is still tracked for the currently running process. Previously if a niced process was added while a non nice process was running the niced process would still get a slice since it was not aware of the unnice process. - Make adjustments for the sched api changes.
2003-04-11 03:47:14 +00:00
kseq = KSEQ_CPU(ke->ke_cpu);
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
kseq_runq_rem(kseq, ke);
kseq_load_rem(kseq, ke);
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
}
fixpt_t
- Update the sched api. sched_{add,rem,clock,pctcpu} now all accept a td argument rather than a kse.
2003-10-16 08:39:15 +00:00
sched_pctcpu(struct thread *td)
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
{
fixpt_t pctcpu;
- Update the sched api. sched_{add,rem,clock,pctcpu} now all accept a td argument rather than a kse.
2003-10-16 08:39:15 +00:00
struct kse *ke;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
pctcpu = 0;
- Update the sched api. sched_{add,rem,clock,pctcpu} now all accept a td argument rather than a kse.
2003-10-16 08:39:15 +00:00
ke = td->td_kse;
- If a thread is not bound to a kse return 0 from sched_pctcpu(). Reported by: pawel.worach@nordea.com
2003-10-20 19:55:21 +00:00
if (ke == NULL)
return (0);
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
- When a new thread is added to a kseq the load is incremented prior to adding it to the nice tables. Therefore, in kseq_add_nice, we should keep in mind that the load will be 1 if we are the only thread, and not 0. - Assert that the sched lock is held in all the appropriate places. - Increase the scope of the sched lock in sched_pctcpu_update(). - Hold the sched lock in sched_runnable(). It is not held by the caller.
2003-06-08 00:47:33 +00:00
mtx_lock_spin(&sched_lock);
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
if (ke->ke_ticks) {
int rtick;
- Fix the maximum slice value. I accidentally checked in a value of '2' which meant no process would run for longer than 20ms. - Slightly redo the interactivity scorer. It follows the same algorithm but in a slightly more correct way. Previously values above half were incorrect. - Lower the interactivity threshold to 20. It seems that in testing non- interactive tasks are hardly ever near there and expensive interactive tasks can sometimes surpass it. This area needs more testing. - Remove an unnecessary KTR. - Fix a case where an idle thread that had an elevated priority due to priority prop. would be placed back on the idle queue. - Delay setting NEEDRESCHED until userret() for threads that haad their priority elevated while in kernel. This gives us the same context switch optimization as SCHED_4BSD. - Limit the child's slice to 1 in sched_fork_kse() so we detect its behavior more quickly. - Inhert some of the run/slp time from the child in sched_exit_ksegrp(). - Redo some of the priority comparisons so they are more clear. - Throttle the frequency of sched_pctcpu_update() so that rounding errors do not make it invalid.
2003-06-15 02:18:29 +00:00
/*
* Don't update more frequently than twice a second. Allowing
* this causes the cpu usage to decay away too quickly due to
* rounding errors.
*/
- Don't let the pctcpu rate limiter throttle us if we have recorded over SCHED_CPU_TICKS ticks. This was allowing processes to display (1/SCHED_CPU_TIME * 100) % more cpu than they had used.
2003-12-11 04:23:39 +00:00
if (ke->ke_ftick + SCHED_CPU_TICKS < ke->ke_ltick ||
ke->ke_ltick < (ticks - (hz / 2)))
- Fix the maximum slice value. I accidentally checked in a value of '2' which meant no process would run for longer than 20ms. - Slightly redo the interactivity scorer. It follows the same algorithm but in a slightly more correct way. Previously values above half were incorrect. - Lower the interactivity threshold to 20. It seems that in testing non- interactive tasks are hardly ever near there and expensive interactive tasks can sometimes surpass it. This area needs more testing. - Remove an unnecessary KTR. - Fix a case where an idle thread that had an elevated priority due to priority prop. would be placed back on the idle queue. - Delay setting NEEDRESCHED until userret() for threads that haad their priority elevated while in kernel. This gives us the same context switch optimization as SCHED_4BSD. - Limit the child's slice to 1 in sched_fork_kse() so we detect its behavior more quickly. - Inhert some of the run/slp time from the child in sched_exit_ksegrp(). - Redo some of the priority comparisons so they are more clear. - Throttle the frequency of sched_pctcpu_update() so that rounding errors do not make it invalid.
2003-06-15 02:18:29 +00:00
sched_pctcpu_update(ke);
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
/* How many rtick per second ? */
- Fix the maximum slice value. I accidentally checked in a value of '2' which meant no process would run for longer than 20ms. - Slightly redo the interactivity scorer. It follows the same algorithm but in a slightly more correct way. Previously values above half were incorrect. - Lower the interactivity threshold to 20. It seems that in testing non- interactive tasks are hardly ever near there and expensive interactive tasks can sometimes surpass it. This area needs more testing. - Remove an unnecessary KTR. - Fix a case where an idle thread that had an elevated priority due to priority prop. would be placed back on the idle queue. - Delay setting NEEDRESCHED until userret() for threads that haad their priority elevated while in kernel. This gives us the same context switch optimization as SCHED_4BSD. - Limit the child's slice to 1 in sched_fork_kse() so we detect its behavior more quickly. - Inhert some of the run/slp time from the child in sched_exit_ksegrp(). - Redo some of the priority comparisons so they are more clear. - Throttle the frequency of sched_pctcpu_update() so that rounding errors do not make it invalid.
2003-06-15 02:18:29 +00:00
rtick = min(ke->ke_ticks / SCHED_CPU_TIME, SCHED_CPU_TICKS);
Use hz if stathz is zero. Adopted from sched_4bsd.
2003-02-02 08:24:32 +00:00
pctcpu = (FSCALE * ((FSCALE * rtick)/realstathz)) >> FSHIFT;
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
}
ke->ke_proc->p_swtime = ke->ke_ltick - ke->ke_ftick;
Protect p_swtime with the sched_lock.
2003-04-22 19:48:25 +00:00
mtx_unlock_spin(&sched_lock);
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
return (pctcpu);
}
- Add initial support for pinning and binding.
2003-11-04 07:45:41 +00:00
void
sched_bind(struct thread *td, int cpu)
{
struct kse *ke;
mtx_assert(&sched_lock, MA_OWNED);
ke = td->td_kse;
ke->ke_flags |= KEF_BOUND;
- Add support for CPU groups to ule. All SMT cores on the same physical cpu are added to a group. - Don't place a cpu into the kseq_idle bitmask until all cpus in that group have idled. - Prefer idle groups over idle group members in the new kseq_transfer() function. In this way we will prefer to balance load across full cores rather than add further load a partial core. - Before a cpu goes idle, check the other group members for threads. Since SMT cpus may freely share threads, this is cheap. - SMT cores may be individually pinned and bound to now. This contrasts the old mechanism where binding or pinning would have allowed a thread to run on any available cpu. - Remove some unnecessary logic from sched_switch(). Priority propagation should be properly taken care of in sched_prio() now.
2003-12-11 03:57:10 +00:00
#ifdef SMP
if (PCPU_GET(cpuid) == cpu)
- Add initial support for pinning and binding.
2003-11-04 07:45:41 +00:00
return;
/* sched_rem without the runq_remove */
ke->ke_state = KES_THREAD;
- Introduce kseq_runq_{add,rem}() which are used to insert and remove kses from the run queues. Also, on SMP, we track the transferable count here. Threads are transferable only as long as they are on the run queue. - Previously, we adjusted our load balancing based on the transferable count minus the number of actual cpus. This was done to account for the threads which were likely to be running. All of this logic is simpler now that transferable accounts for only those threads which can actually be taken. Updated various places in sched_add() and kseq_balance() to account for this. - Rename kseq_{add,rem} to kseq_load_{add,rem} to reflect what they're really doing. The load is accounted for seperately from the runq because the load is accounted for even as the thread is running. - Fix a bug in sched_class() where we weren't properly using the PRI_BASE() version of the kg_pri_class. - Add a large comment that describes the impact of a seemingly simple conditional in sched_add(). - Also in sched_add() check the transferable count and KSE_CAN_MIGRATE() prior to checking kseq_idle. This reduces the frequency of access for kseq_idle which is a shared resource.
2003-11-15 07:32:07 +00:00
kseq_load_rem(KSEQ_CPU(ke->ke_cpu), ke);
- Add initial support for pinning and binding.
2003-11-04 07:45:41 +00:00
kseq_notify(ke, cpu);
/* When we return from mi_switch we'll be on the correct cpu. */
Add NULL arg to mi_switch() call to stop kernel compiles from breaking.
2004-07-03 16:57:51 +00:00
mi_switch(SW_VOL, NULL);
- Add initial support for pinning and binding.
2003-11-04 07:45:41 +00:00
#endif
}
void
sched_unbind(struct thread *td)
{
mtx_assert(&sched_lock, MA_OWNED);
td->td_kse->ke_flags &= ~KEF_BOUND;
}
- Add a new member to struct kseq called ksq_sysload. This is intended to track the load for the sched_load() function. In the SMP case this member is not defined because it would be redundant with the ksg_load member which already tracks the non ithd load. - For sched_load() in the UP case simply return ksq_sysload. In the SMP case traverse the list of kseq groups and sum up their ksg_load fields.
2004-02-01 02:48:36 +00:00
int
sched_load(void)
{
#ifdef SMP
int total;
int i;
total = 0;
for (i = 0; i <= ksg_maxid; i++)
total += KSEQ_GROUP(i)->ksg_load;
return (total);
#else
return (KSEQ_SELF()->ksq_sysload);
#endif
}
- Add the ule scheduler. This is intended to be a general purpose process scheduler with many SMP benefits. It is still very experimental and should be used only in test environments.
2003-01-26 05:23:15 +00:00
int
sched_sizeof_ksegrp(void)
{
return (sizeof(struct ksegrp) + sizeof(struct kg_sched));
}
int
sched_sizeof_proc(void)
{
return (sizeof(struct proc));
}
int
sched_sizeof_thread(void)
{
return (sizeof(struct thread) + sizeof(struct td_sched));
}
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
#define KERN_SWITCH_INCLUDE 1
#include "kern/kern_switch.c"
Reference in New Issue Copy Permalink