- 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.
This commit is contained in:
Jeff Roberson 2003-06-15 02:18:29 +00:00
parent 1356c67ebe
commit 210491d3d9

View File

@ -69,7 +69,7 @@ SYSCTL_INT(_kern_sched, OID_AUTO, strict, CTLFLAG_RD, &sched_strict, 0, "");
static int slice_min = 1;
SYSCTL_INT(_kern_sched, OID_AUTO, slice_min, CTLFLAG_RW, &slice_min, 0, "");
static int slice_max = 2;
static int slice_max = 10;
SYSCTL_INT(_kern_sched, OID_AUTO, slice_max, CTLFLAG_RW, &slice_max, 0, "");
int realstathz;
@ -140,7 +140,7 @@ struct td_sched *thread0_sched = &td_sched;
#define SCHED_PRI_BASE ((SCHED_PRI_NRESV / 2) + PRI_MIN_TIMESHARE)
#define SCHED_DYN_RANGE (SCHED_PRI_RANGE - SCHED_PRI_NRESV)
#define SCHED_PRI_INTERACT(score) \
((score) * SCHED_DYN_RANGE / SCHED_INTERACT_RANGE)
((score) * SCHED_DYN_RANGE / SCHED_INTERACT_MAX)
/*
* These determine the interactivity of a process.
@ -148,15 +148,14 @@ struct td_sched *thread0_sched = &td_sched;
* SLP_RUN_MAX: Maximum amount of sleep time + run time we'll accumulate
* before throttling back.
* SLP_RUN_THROTTLE: Divisor for reducing slp/run time.
* INTERACT_RANGE: Range of interactivity values. Smaller is better.
* INTERACT_HALF: Convenience define, half of the interactivity range.
* INTERACT_MAX: Maximum interactivity value. Smaller is better.
* INTERACT_THRESH: Threshhold for placement on the current runq.
*/
#define SCHED_SLP_RUN_MAX ((hz / 10) << 10)
#define SCHED_SLP_RUN_THROTTLE (10)
#define SCHED_INTERACT_RANGE (100)
#define SCHED_INTERACT_HALF (SCHED_INTERACT_RANGE / 2)
#define SCHED_INTERACT_THRESH (10)
#define SCHED_INTERACT_MAX (100)
#define SCHED_INTERACT_HALF (SCHED_INTERACT_MAX / 2)
#define SCHED_INTERACT_THRESH (20)
/*
* These parameters and macros determine the size of the time slice that is
@ -619,10 +618,6 @@ sched_slice(struct kse *ke)
* in the kg. This will cause us to forget old interactivity
* while maintaining the current ratio.
*/
CTR4(KTR_ULE, "Slp vs Run %p (Slp %d, Run %d, Score %d)",
ke, kg->kg_slptime >> 10, kg->kg_runtime >> 10,
sched_interact_score(kg));
if ((kg->kg_runtime + kg->kg_slptime) > SCHED_SLP_RUN_MAX) {
kg->kg_runtime /= SCHED_SLP_RUN_THROTTLE;
kg->kg_slptime /= SCHED_SLP_RUN_THROTTLE;
@ -637,29 +632,22 @@ sched_slice(struct kse *ke)
static int
sched_interact_score(struct ksegrp *kg)
{
int big;
int small;
int base;
int div;
if (kg->kg_runtime > kg->kg_slptime) {
big = kg->kg_runtime;
small = kg->kg_slptime;
base = SCHED_INTERACT_HALF;
} else {
big = kg->kg_slptime;
small = kg->kg_runtime;
base = 0;
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);
}
big /= SCHED_INTERACT_HALF;
if (big != 0)
small /= big;
else
small = 0;
/*
* This can happen if slptime and runtime are 0.
*/
return (0);
small += base;
/* XXX Factor in nice */
return (small);
}
/*
@ -678,7 +666,9 @@ sched_pctcpu_update(struct kse *ke)
{
/*
* Adjust counters and watermark for pctcpu calc.
*
*/
/*
* Shift the tick count out so that the divide doesn't round away
* our results.
*/
@ -761,6 +751,13 @@ sched_switchout(struct thread *td)
td->td_flags &= ~TDF_NEEDRESCHED;
if (TD_IS_RUNNING(td)) {
/*
* This queue is always correct except for idle threads which
* have a higher priority due to priority propagation.
*/
if (ke->ke_ksegrp->kg_pri_class == PRI_IDLE &&
ke->ke_thread->td_priority > PRI_MIN_IDLE)
ke->ke_runq = KSEQ_SELF()->ksq_curr;
runq_add(ke->ke_runq, ke);
/* setrunqueue(td); */
return;
@ -782,10 +779,6 @@ sched_switchin(struct thread *td)
mtx_assert(&sched_lock, MA_OWNED);
td->td_oncpu = PCPU_GET(cpuid);
if (td->td_ksegrp->kg_pri_class == PRI_TIMESHARE &&
td->td_priority != td->td_ksegrp->kg_user_pri)
curthread->td_flags |= TDF_NEEDRESCHED;
}
void
@ -872,7 +865,7 @@ void
sched_fork_kse(struct kse *ke, struct kse *child)
{
child->ke_slice = ke->ke_slice;
child->ke_slice = 1; /* Attempt to quickly learn interactivity. */
child->ke_cpu = ke->ke_cpu; /* sched_pickcpu(); */
child->ke_runq = NULL;
@ -891,6 +884,11 @@ sched_fork_ksegrp(struct ksegrp *kg, struct ksegrp *child)
PROC_LOCK_ASSERT(child->kg_proc, MA_OWNED);
/* XXX Need something better here */
#if 1
child->kg_slptime = kg->kg_slptime;
child->kg_runtime = kg->kg_runtime;
#else
if (kg->kg_slptime > kg->kg_runtime) {
child->kg_slptime = SCHED_DYN_RANGE;
child->kg_runtime = kg->kg_slptime / SCHED_DYN_RANGE;
@ -898,6 +896,7 @@ sched_fork_ksegrp(struct ksegrp *kg, struct ksegrp *child)
child->kg_runtime = SCHED_DYN_RANGE;
child->kg_slptime = kg->kg_runtime / SCHED_DYN_RANGE;
}
#endif
child->kg_user_pri = kg->kg_user_pri;
child->kg_nice = kg->kg_nice;
@ -945,6 +944,7 @@ sched_exit(struct proc *p, struct proc *child)
/* XXX Need something better here */
mtx_assert(&sched_lock, MA_OWNED);
sched_exit_kse(FIRST_KSE_IN_PROC(p), FIRST_KSE_IN_PROC(child));
sched_exit_ksegrp(FIRST_KSEGRP_IN_PROC(p), FIRST_KSEGRP_IN_PROC(child));
}
void
@ -956,6 +956,8 @@ sched_exit_kse(struct kse *ke, struct kse *child)
void
sched_exit_ksegrp(struct ksegrp *kg, struct ksegrp *child)
{
kg->kg_slptime += child->kg_slptime;
kg->kg_runtime += child->kg_runtime;
}
void
@ -1101,12 +1103,20 @@ void
sched_userret(struct thread *td)
{
struct ksegrp *kg;
struct kseq *kseq;
struct kse *ke;
kg = td->td_ksegrp;
if (td->td_priority != kg->kg_user_pri) {
mtx_lock_spin(&sched_lock);
td->td_priority = kg->kg_user_pri;
kseq = KSEQ_SELF();
if (td->td_ksegrp->kg_pri_class == PRI_TIMESHARE &&
kseq->ksq_load > 1 &&
(ke = kseq_choose(kseq)) != NULL &&
ke->ke_thread->td_priority < td->td_priority)
curthread->td_flags |= TDF_NEEDRESCHED;
mtx_unlock_spin(&sched_lock);
}
}
@ -1196,7 +1206,7 @@ sched_add(struct kse *ke)
/*
* This is for priority prop.
*/
if (ke->ke_thread->td_priority < PRI_MAX_TIMESHARE)
if (ke->ke_thread->td_priority > PRI_MIN_IDLE)
ke->ke_runq = kseq->ksq_curr;
else
ke->ke_runq = &kseq->ksq_idle;
@ -1242,10 +1252,16 @@ sched_pctcpu(struct kse *ke)
/* Update to account for time potentially spent sleeping */
ke->ke_ltick = ticks;
/*
* Don't update more frequently than twice a second. Allowing
* this causes the cpu usage to decay away too quickly due to
* rounding errors.
*/
if (ke->ke_ltick < (ticks - (hz / 2)))
sched_pctcpu_update(ke);
/* How many rtick per second ? */
rtick = ke->ke_ticks / SCHED_CPU_TIME;
rtick = min(ke->ke_ticks / SCHED_CPU_TIME, SCHED_CPU_TICKS);
pctcpu = (FSCALE * ((FSCALE * rtick)/realstathz)) >> FSHIFT;
}