1
0
mirror of https://git.FreeBSD.org/src.git synced 2024-12-01 08:27:59 +00:00

Don't release and acquire Giant in mi_switch(). Instead, release and

acquire Giant as needed in functions that call mi_switch().  The releases
need to be done outside of the sched_lock to avoid potential deadlocks
from trying to acquire Giant while interrupts are disabled.

Submitted by:	witness
This commit is contained in:
John Baldwin 2000-11-16 02:16:44 +00:00
parent 088638dae4
commit 20cdcc5b73
Notes: svn2git 2020-12-20 02:59:44 +00:00
svn path=/head/; revision=68808
12 changed files with 39 additions and 133 deletions

View File

@ -120,11 +120,13 @@ userret(p, pc, oticks, have_giant)
* indicated by our priority.
*/
s = splstatclock();
DROP_GIANT_NOSWITCH();
mtx_enter(&sched_lock, MTX_SPIN);
setrunqueue(p);
p->p_stats->p_ru.ru_nivcsw++;
mi_switch();
mtx_exit(&sched_lock, MTX_SPIN);
PICKUP_GIANT();
splx(s);
while ((sig = CURSIG(p)) != 0) {
if (have_giant == 0) {

View File

@ -192,11 +192,13 @@ userret(p, frame, oticks, have_giant)
* our priority.
*/
s = splhigh();
DROP_GIANT_NOSWITCH();
mtx_enter(&sched_lock, MTX_SPIN);
setrunqueue(p);
p->p_stats->p_ru.ru_nivcsw++;
mi_switch();
mtx_exit(&sched_lock, MTX_SPIN);
PICKUP_GIANT();
splx(s);
while ((sig = CURSIG(p)) != 0) {
if (have_giant == 0) {

View File

@ -192,11 +192,13 @@ userret(p, frame, oticks, have_giant)
* our priority.
*/
s = splhigh();
DROP_GIANT_NOSWITCH();
mtx_enter(&sched_lock, MTX_SPIN);
setrunqueue(p);
p->p_stats->p_ru.ru_nivcsw++;
mi_switch();
mtx_exit(&sched_lock, MTX_SPIN);
PICKUP_GIANT();
splx(s);
while ((sig = CURSIG(p)) != 0) {
if (have_giant == 0) {

View File

@ -102,11 +102,13 @@ userret(register struct proc *p, u_int64_t pc, u_quad_t oticks, int have_giant)
* indicated by our priority.
*/
s = splstatclock();
DROP_GIANT_NOSWITCH();
mtx_enter(&sched_lock, MTX_SPIN);
setrunqueue(p);
p->p_stats->p_ru.ru_nivcsw++;
mi_switch();
mtx_exit(&sched_lock, MTX_SPIN);
PICKUP_GIANT();
splx(s);
while ((sig = CURSIG(p)) != 0) {
if (have_giant == 0) {

View File

@ -223,7 +223,6 @@ void
mtx_enter_hard(struct mtx *m, int type, int saveintr)
{
struct proc *p = CURPROC;
struct timeval new_switchtime;
KASSERT(p != NULL, ("curproc is NULL in mutex"));
@ -323,42 +322,7 @@ mtx_enter_hard(struct mtx *m, int type, int saveintr)
#endif
CTR3(KTR_LOCK, "mtx_enter: p 0x%p blocked on [0x%p] %s",
p, m, m->mtx_description);
/*
* Blatantly copied from mi_switch nearly verbatim.
* When Giant goes away and we stop dinking with it
* in mi_switch, we can go back to calling mi_switch
* directly here.
*/
/*
* Compute the amount of time during which the current
* process was running, and add that to its total so
* far.
*/
microuptime(&new_switchtime);
if (timevalcmp(&new_switchtime, &switchtime, <)) {
printf(
"microuptime() went backwards (%ld.%06ld -> %ld.%06ld)\n",
switchtime.tv_sec, switchtime.tv_usec,
new_switchtime.tv_sec,
new_switchtime.tv_usec);
new_switchtime = switchtime;
} else {
p->p_runtime += (new_switchtime.tv_usec -
switchtime.tv_usec) +
(new_switchtime.tv_sec - switchtime.tv_sec) *
(int64_t)1000000;
}
/*
* Pick a new current process and record its start time.
*/
cnt.v_swtch++;
switchtime = new_switchtime;
cpu_switch();
if (switchtime.tv_sec == 0)
microuptime(&switchtime);
switchticks = ticks;
mi_switch();
CTR3(KTR_LOCK,
"mtx_enter: p 0x%p free from blocked on [0x%p] %s",
p, m, m->mtx_description);
@ -735,8 +699,10 @@ static char *ignore_list[] = {
static char *spin_order_list[] = {
"sched lock",
"clk",
"sio",
#ifdef __i386__
"clk",
#endif
/*
* leaf locks
*/
@ -752,7 +718,6 @@ static char *dup_list[] = {
};
static char *sleep_list[] = {
"Giant lock",
NULL
};

View File

@ -254,6 +254,7 @@ boot(int howto)
pbusy = nbusy;
sync(&proc0, NULL);
if (curproc != NULL) {
DROP_GIANT_NOSWITCH();
for (subiter = 0; subiter < 50 * iter; subiter++) {
mtx_enter(&sched_lock, MTX_SPIN);
setrunqueue(curproc);
@ -261,6 +262,7 @@ boot(int howto)
mtx_exit(&sched_lock, MTX_SPIN);
DELAY(1000);
}
PICKUP_GIANT();
} else
DELAY(50000 * iter);
}

View File

@ -1283,7 +1283,11 @@ issignal(p)
psignal(p->p_pptr, SIGCHLD);
do {
stop(p);
DROP_GIANT_NOSWITCH();
mtx_enter(&sched_lock, MTX_SPIN);
mi_switch();
mtx_exit(&sched_lock, MTX_SPIN);
PICKUP_GIANT();
} while (!trace_req(p)
&& p->p_flag & P_TRACED);
@ -1352,7 +1356,11 @@ issignal(p)
stop(p);
if ((p->p_pptr->p_procsig->ps_flag & PS_NOCLDSTOP) == 0)
psignal(p->p_pptr, SIGCHLD);
DROP_GIANT_NOSWITCH();
mtx_enter(&sched_lock, MTX_SPIN);
mi_switch();
mtx_exit(&sched_lock, MTX_SPIN);
PICKUP_GIANT();
break;
} else if (prop & SA_IGNORE) {
/*

View File

@ -377,11 +377,13 @@ uio_yield()
p = curproc;
s = splhigh();
DROP_GIANT_NOSWITCH();
mtx_enter(&sched_lock, MTX_SPIN);
p->p_priority = p->p_usrpri;
setrunqueue(p);
p->p_stats->p_ru.ru_nivcsw++;
mi_switch();
mtx_exit(&sched_lock, MTX_SPIN);
PICKUP_GIANT();
splx(s);
}

View File

@ -420,6 +420,7 @@ msleep(ident, mtx, priority, wmesg, timo)
if (p && KTRPOINT(p, KTR_CSW))
ktrcsw(p->p_tracep, 1, 0);
#endif
DROP_GIANT_NOSWITCH();
WITNESS_SLEEP(0, mtx);
mtx_enter(&sched_lock, MTX_SPIN);
@ -533,6 +534,7 @@ msleep(ident, mtx, priority, wmesg, timo)
if (KTRPOINT(p, KTR_CSW))
ktrcsw(p->p_tracep, 0, 0);
#endif
PICKUP_GIANT();
if (mtx != NULL) {
mtx_enter(mtx, MTX_DEF);
WITNESS_RESTORE(mtx, mtx);
@ -610,6 +612,7 @@ mawait(struct mtx *mtx, int priority, int timo)
int s;
WITNESS_SAVE_DECL(mtx);
DROP_GIANT_NOSWITCH();
WITNESS_SLEEP(0, mtx);
mtx_enter(&sched_lock, MTX_SPIN);
if (mtx != NULL) {
@ -724,6 +727,7 @@ mawait(struct mtx *mtx, int priority, int timo)
p->p_asleep.as_priority = 0;
out:
PICKUP_GIANT();
if (mtx != NULL) {
mtx_enter(mtx, MTX_DEF);
WITNESS_RESTORE(mtx, mtx);
@ -881,9 +885,7 @@ mi_switch()
struct timeval new_switchtime;
register struct proc *p = curproc; /* XXX */
register struct rlimit *rlim;
int giantreleased;
int x;
WITNESS_SAVE_DECL(Giant);
/*
* XXX this spl is almost unnecessary. It is partly to allow for
@ -904,14 +906,7 @@ mi_switch()
*/
x = splstatclock();
CTR4(KTR_PROC, "mi_switch: old proc %p (pid %d, %s), schedlock %p",
p, p->p_pid, p->p_comm, (void *) sched_lock.mtx_lock);
mtx_enter(&sched_lock, MTX_SPIN | MTX_RLIKELY);
if (mtx_owned(&Giant))
WITNESS_SAVE(&Giant, Giant);
for (giantreleased = 0; mtx_owned(&Giant); giantreleased++)
mtx_exit(&Giant, MTX_DEF | MTX_NOSWITCH);
mtx_assert(&sched_lock, MA_OWNED);
#ifdef SIMPLELOCK_DEBUG
if (p->p_simple_locks)
@ -965,12 +960,6 @@ mi_switch()
if (switchtime.tv_sec == 0)
microuptime(&switchtime);
switchticks = ticks;
mtx_exit(&sched_lock, MTX_SPIN);
while (giantreleased--)
mtx_enter(&Giant, MTX_DEF);
if (mtx_owned(&Giant))
WITNESS_RESTORE(&Giant, Giant);
splx(x);
}

View File

@ -192,11 +192,13 @@ userret(p, frame, oticks, have_giant)
* our priority.
*/
s = splhigh();
DROP_GIANT_NOSWITCH();
mtx_enter(&sched_lock, MTX_SPIN);
setrunqueue(p);
p->p_stats->p_ru.ru_nivcsw++;
mi_switch();
mtx_exit(&sched_lock, MTX_SPIN);
PICKUP_GIANT();
splx(s);
while ((sig = CURSIG(p)) != 0) {
if (have_giant == 0) {

View File

@ -223,7 +223,6 @@ void
mtx_enter_hard(struct mtx *m, int type, int saveintr)
{
struct proc *p = CURPROC;
struct timeval new_switchtime;
KASSERT(p != NULL, ("curproc is NULL in mutex"));
@ -323,42 +322,7 @@ mtx_enter_hard(struct mtx *m, int type, int saveintr)
#endif
CTR3(KTR_LOCK, "mtx_enter: p 0x%p blocked on [0x%p] %s",
p, m, m->mtx_description);
/*
* Blatantly copied from mi_switch nearly verbatim.
* When Giant goes away and we stop dinking with it
* in mi_switch, we can go back to calling mi_switch
* directly here.
*/
/*
* Compute the amount of time during which the current
* process was running, and add that to its total so
* far.
*/
microuptime(&new_switchtime);
if (timevalcmp(&new_switchtime, &switchtime, <)) {
printf(
"microuptime() went backwards (%ld.%06ld -> %ld.%06ld)\n",
switchtime.tv_sec, switchtime.tv_usec,
new_switchtime.tv_sec,
new_switchtime.tv_usec);
new_switchtime = switchtime;
} else {
p->p_runtime += (new_switchtime.tv_usec -
switchtime.tv_usec) +
(new_switchtime.tv_sec - switchtime.tv_sec) *
(int64_t)1000000;
}
/*
* Pick a new current process and record its start time.
*/
cnt.v_swtch++;
switchtime = new_switchtime;
cpu_switch();
if (switchtime.tv_sec == 0)
microuptime(&switchtime);
switchticks = ticks;
mi_switch();
CTR3(KTR_LOCK,
"mtx_enter: p 0x%p free from blocked on [0x%p] %s",
p, m, m->mtx_description);
@ -735,8 +699,10 @@ static char *ignore_list[] = {
static char *spin_order_list[] = {
"sched lock",
"clk",
"sio",
#ifdef __i386__
"clk",
#endif
/*
* leaf locks
*/
@ -752,7 +718,6 @@ static char *dup_list[] = {
};
static char *sleep_list[] = {
"Giant lock",
NULL
};

View File

@ -223,7 +223,6 @@ void
mtx_enter_hard(struct mtx *m, int type, int saveintr)
{
struct proc *p = CURPROC;
struct timeval new_switchtime;
KASSERT(p != NULL, ("curproc is NULL in mutex"));
@ -323,42 +322,7 @@ mtx_enter_hard(struct mtx *m, int type, int saveintr)
#endif
CTR3(KTR_LOCK, "mtx_enter: p 0x%p blocked on [0x%p] %s",
p, m, m->mtx_description);
/*
* Blatantly copied from mi_switch nearly verbatim.
* When Giant goes away and we stop dinking with it
* in mi_switch, we can go back to calling mi_switch
* directly here.
*/
/*
* Compute the amount of time during which the current
* process was running, and add that to its total so
* far.
*/
microuptime(&new_switchtime);
if (timevalcmp(&new_switchtime, &switchtime, <)) {
printf(
"microuptime() went backwards (%ld.%06ld -> %ld.%06ld)\n",
switchtime.tv_sec, switchtime.tv_usec,
new_switchtime.tv_sec,
new_switchtime.tv_usec);
new_switchtime = switchtime;
} else {
p->p_runtime += (new_switchtime.tv_usec -
switchtime.tv_usec) +
(new_switchtime.tv_sec - switchtime.tv_sec) *
(int64_t)1000000;
}
/*
* Pick a new current process and record its start time.
*/
cnt.v_swtch++;
switchtime = new_switchtime;
cpu_switch();
if (switchtime.tv_sec == 0)
microuptime(&switchtime);
switchticks = ticks;
mi_switch();
CTR3(KTR_LOCK,
"mtx_enter: p 0x%p free from blocked on [0x%p] %s",
p, m, m->mtx_description);
@ -735,8 +699,10 @@ static char *ignore_list[] = {
static char *spin_order_list[] = {
"sched lock",
"clk",
"sio",
#ifdef __i386__
"clk",
#endif
/*
* leaf locks
*/
@ -752,7 +718,6 @@ static char *dup_list[] = {
};
static char *sleep_list[] = {
"Giant lock",
NULL
};