/*
* Layout:
* - column counts
* - header
* - single-threaded process
* - multi-threaded process
* - thread in a MT process
*
* 1 2 3 4 5 6 7
* 1234567890123456789012345678901234567890123456789012345678901234567890
* pid ppid pgrp uid state wmesg wchan cmd
* <pid> <ppi> <pgi> <uid> <stat> < wmesg > < wchan > <name>
* <pid> <ppi> <pgi> <uid> <stat> (threaded) <command>
* <tid > <stat> < wmesg > < wchan > <name>
*
* For machines with 64-bit pointers, we expand the wchan field 8 more
* characters.
*/
void
db_ps(db_expr_t addr, boolean_t hasaddr, db_expr_t count, char *modif)
{
volatile struct proc *p, *pp;
volatile struct thread *td;
struct ucred *cred;
struct pgrp *pgrp;
char state[9];
int np, rflag, sflag, dflag, lflag, wflag;
np = nprocs;
if (!LIST_EMPTY(&allproc))
p = LIST_FIRST(&allproc);
else
p = &proc0;
#ifdef __LP64__
db_printf(" pid ppid pgrp uid state wmesg wchan cmd\n");
#else
db_printf(" pid ppid pgrp uid state wmesg wchan cmd\n");
#endif
while (--np >= 0 && !db_pager_quit) {
if (p == NULL) {
db_printf("oops, ran out of processes early!\n");
break;
}
pp = p->p_pptr;
if (pp == NULL)
pp = p;
cred = p->p_ucred;
pgrp = p->p_pgrp;
db_printf("%5d %5d %5d %5d ", p->p_pid, pp->p_pid,
pgrp != NULL ? pgrp->pg_id : 0,
cred != NULL ? cred->cr_ruid : 0);
/* Determine our primary process state. */
switch (p->p_state) {
case PRS_NORMAL:
if (P_SHOULDSTOP(p))
state[0] = 'T';
else {
/*
* One of D, L, R, S, W. For a
* multithreaded process we will use
* the state of the thread with the
* highest precedence. The
* precendence order from high to low
* is R, L, D, S, W. If no thread is
* in a sane state we use '?' for our
* primary state.
*/
rflag = sflag = dflag = lflag = wflag = 0;
FOREACH_THREAD_IN_PROC(p, td) {
if (td->td_state == TDS_RUNNING ||
td->td_state == TDS_RUNQ ||
td->td_state == TDS_CAN_RUN)
rflag++;
if (TD_ON_LOCK(td))
lflag++;
if (TD_IS_SLEEPING(td)) {
if (!(td->td_flags & TDF_SINTR))
dflag++;
else
sflag++;
}
if (TD_AWAITING_INTR(td))
wflag++;
}
if (rflag)
state[0] = 'R';
else if (lflag)
state[0] = 'L';
else if (dflag)
state[0] = 'D';
else if (sflag)
state[0] = 'S';
else if (wflag)
state[0] = 'W';
else
state[0] = '?';
}
break;
case PRS_NEW:
state[0] = 'N';
break;
case PRS_ZOMBIE:
state[0] = 'Z';
break;
default:
state[0] = 'U';
break;
}
state[1] = '\0';
/* Additional process state flags. */
if (!(p->p_flag & P_INMEM))
strlcat(state, "W", sizeof(state));
if (p->p_flag & P_TRACED)
strlcat(state, "X", sizeof(state));
if (p->p_flag & P_WEXIT && p->p_state != PRS_ZOMBIE)
strlcat(state, "E", sizeof(state));
if (p->p_flag & P_PPWAIT)
strlcat(state, "V", sizeof(state));
if (p->p_flag & P_SYSTEM || p->p_lock > 0)
strlcat(state, "L", sizeof(state));
if (p->p_session != NULL && SESS_LEADER(p))
strlcat(state, "s", sizeof(state));
/* Cheated here and didn't compare pgid's. */
if (p->p_flag & P_CONTROLT)
strlcat(state, "+", sizeof(state));
if (cred != NULL && jailed(cred))
strlcat(state, "J", sizeof(state));
db_printf(" %-6.6s ", state);
if (p->p_flag & P_HADTHREADS) {
#ifdef __LP64__
db_printf(" (threaded) ");
#else
db_printf(" (threaded) ");
#endif
if (p->p_flag & P_SYSTEM)
db_printf("[");
db_printf("%s", p->p_comm);
if (p->p_flag & P_SYSTEM)
db_printf("]");
db_printf("\n");
}
FOREACH_THREAD_IN_PROC(p, td) {
dumpthread(p, td, p->p_flag & P_HADTHREADS);
if (db_pager_quit)
break;
}
p = LIST_NEXT(p, p_list);
if (p == NULL && np > 0)
p = LIST_FIRST(&zombproc);
}
/*
* Original vm_pageout_oom, will be called if LRU pageout_oom will fail
*/
static void
original_vm_pageout_oom(int shortage)
{
struct proc *p, *bigproc;
vm_offset_t size, bigsize;
struct thread *td;
struct vmspace *vm;
/*
* We keep the process bigproc locked once we find it to keep anyone
* from messing with it; however, there is a possibility of
* deadlock if process B is bigproc and one of it's child processes
* attempts to propagate a signal to B while we are waiting for A's
* lock while walking this list. To avoid this, we don't block on
* the process lock but just skip a process if it is already locked.
*/
bigproc = NULL;
bigsize = 0;
sx_slock(&allproc_lock);
FOREACH_PROC_IN_SYSTEM(p) {
int breakout;
if (PROC_TRYLOCK(p) == 0)
continue;
/*
* If this is a system, protected or killed process, skip it.
*/
if (p->p_state != PRS_NORMAL ||
(p->p_flag & (P_INEXEC | P_PROTECTED | P_SYSTEM)) ||
(p->p_pid == 1) || P_KILLED(p) ||
((p->p_pid < 48) && (swap_pager_avail != 0))) {
PROC_UNLOCK(p);
continue;
}
/*
* If the process is in a non-running type state,
* don't touch it. Check all the threads individually.
*/
breakout = 0;
FOREACH_THREAD_IN_PROC(p, td) {
thread_lock(td);
if (!TD_ON_RUNQ(td) &&
!TD_IS_RUNNING(td) &&
!TD_IS_SLEEPING(td)) {
thread_unlock(td);
breakout = 1;
break;
}
thread_unlock(td);
}
if (breakout) {
PROC_UNLOCK(p);
continue;
}
/*
* get the process size
*/
vm = vmspace_acquire_ref(p);
if (vm == NULL) {
PROC_UNLOCK(p);
continue;
}
if (!vm_map_trylock_read(&vm->vm_map)) {
vmspace_free(vm);
PROC_UNLOCK(p);
continue;
}
size = vmspace_swap_count(vm);
vm_map_unlock_read(&vm->vm_map);
if (shortage == VM_OOM_MEM)
size += vmspace_resident_count(vm);
vmspace_free(vm);
/*
* if the this process is bigger than the biggest one
* remember it.
*/
if (size > bigsize) {
if (bigproc != NULL)
PROC_UNLOCK(bigproc);
bigproc = p;
bigsize = size;
} else
PROC_UNLOCK(p);
}
static void
deadlkres(void)
{
struct proc *p;
struct thread *td;
void *wchan;
int blkticks, i, slpticks, slptype, tryl, tticks;
tryl = 0;
for (;;) {
blkticks = blktime_threshold * hz;
slpticks = slptime_threshold * hz;
/*
* Avoid to sleep on the sx_lock in order to avoid a possible
* priority inversion problem leading to starvation.
* If the lock can't be held after 100 tries, panic.
*/
if (!sx_try_slock(&allproc_lock)) {
if (tryl > 100)
panic("%s: possible deadlock detected on allproc_lock\n",
__func__);
tryl++;
pause("allproc", sleepfreq * hz);
continue;
}
tryl = 0;
FOREACH_PROC_IN_SYSTEM(p) {
PROC_LOCK(p);
if (p->p_state == PRS_NEW) {
PROC_UNLOCK(p);
continue;
}
FOREACH_THREAD_IN_PROC(p, td) {
/*
* Once a thread is found in "interesting"
* state a possible ticks wrap-up needs to be
* checked.
*/
thread_lock(td);
if (TD_ON_LOCK(td) && ticks < td->td_blktick) {
/*
* The thread should be blocked on a
* turnstile, simply check if the
* turnstile channel is in good state.
*/
MPASS(td->td_blocked != NULL);
tticks = ticks - td->td_blktick;
thread_unlock(td);
if (tticks > blkticks) {
/*
* Accordingly with provided
* thresholds, this thread is
* stuck for too long on a
* turnstile.
*/
PROC_UNLOCK(p);
sx_sunlock(&allproc_lock);
panic("%s: possible deadlock detected for %p, blocked for %d ticks\n",
__func__, td, tticks);
}
} else if (TD_IS_SLEEPING(td) &&
TD_ON_SLEEPQ(td) &&
ticks < td->td_blktick) {
/*
* Check if the thread is sleeping on a
* lock, otherwise skip the check.
* Drop the thread lock in order to
* avoid a LOR with the sleepqueue
* spinlock.
*/
wchan = td->td_wchan;
tticks = ticks - td->td_slptick;
thread_unlock(td);
slptype = sleepq_type(wchan);
if ((slptype == SLEEPQ_SX ||
slptype == SLEEPQ_LK) &&
tticks > slpticks) {
/*
* Accordingly with provided
* thresholds, this thread is
* stuck for too long on a
* sleepqueue.
* However, being on a
* sleepqueue, we might still
* check for the blessed
* list.
*/
tryl = 0;
for (i = 0; blessed[i] != NULL;
i++) {
if (!strcmp(blessed[i],
td->td_wmesg)) {
tryl = 1;
break;
}
}
if (tryl != 0) {
tryl = 0;
continue;
}
PROC_UNLOCK(p);
sx_sunlock(&allproc_lock);
panic("%s: possible deadlock detected for %p, blocked for %d ticks\n",
__func__, td, tticks);
}
} else
thread_unlock(td);
}
PROC_UNLOCK(p);
}
sx_sunlock(&allproc_lock);
/* Sleep for sleepfreq seconds. */
pause("-", sleepfreq * hz);
}
/*
* Read proc's from memory file into buffer bp, which has space to hold
* at most maxcnt procs.
*/
static int
kvm_proclist(kvm_t *kd, int what, int arg, struct proc *p,
struct kinfo_proc *bp, int maxcnt)
{
int cnt = 0;
struct kinfo_proc kinfo_proc, *kp;
struct pgrp pgrp;
struct session sess;
struct cdev t_cdev;
struct tty tty;
struct vmspace vmspace;
struct sigacts sigacts;
#if 0
struct pstats pstats;
#endif
struct ucred ucred;
struct prison pr;
struct thread mtd;
struct proc proc;
struct proc pproc;
struct sysentvec sysent;
char svname[KI_EMULNAMELEN];
kp = &kinfo_proc;
kp->ki_structsize = sizeof(kinfo_proc);
/*
* Loop on the processes. this is completely broken because we need to be
* able to loop on the threads and merge the ones that are the same process some how.
*/
for (; cnt < maxcnt && p != NULL; p = LIST_NEXT(&proc, p_list)) {
memset(kp, 0, sizeof *kp);
if (KREAD(kd, (u_long)p, &proc)) {
_kvm_err(kd, kd->program, "can't read proc at %p", p);
return (-1);
}
if (proc.p_state == PRS_NEW)
continue;
if (proc.p_state != PRS_ZOMBIE) {
if (KREAD(kd, (u_long)TAILQ_FIRST(&proc.p_threads),
&mtd)) {
_kvm_err(kd, kd->program,
"can't read thread at %p",
TAILQ_FIRST(&proc.p_threads));
return (-1);
}
}
if (KREAD(kd, (u_long)proc.p_ucred, &ucred) == 0) {
kp->ki_ruid = ucred.cr_ruid;
kp->ki_svuid = ucred.cr_svuid;
kp->ki_rgid = ucred.cr_rgid;
kp->ki_svgid = ucred.cr_svgid;
kp->ki_cr_flags = ucred.cr_flags;
if (ucred.cr_ngroups > KI_NGROUPS) {
kp->ki_ngroups = KI_NGROUPS;
kp->ki_cr_flags |= KI_CRF_GRP_OVERFLOW;
} else
kp->ki_ngroups = ucred.cr_ngroups;
kvm_read(kd, (u_long)ucred.cr_groups, kp->ki_groups,
kp->ki_ngroups * sizeof(gid_t));
kp->ki_uid = ucred.cr_uid;
if (ucred.cr_prison != NULL) {
if (KREAD(kd, (u_long)ucred.cr_prison, &pr)) {
_kvm_err(kd, kd->program,
"can't read prison at %p",
ucred.cr_prison);
return (-1);
}
kp->ki_jid = pr.pr_id;
}
}
switch(what & ~KERN_PROC_INC_THREAD) {
case KERN_PROC_GID:
if (kp->ki_groups[0] != (gid_t)arg)
continue;
break;
case KERN_PROC_PID:
if (proc.p_pid != (pid_t)arg)
continue;
break;
case KERN_PROC_RGID:
if (kp->ki_rgid != (gid_t)arg)
continue;
break;
case KERN_PROC_UID:
if (kp->ki_uid != (uid_t)arg)
continue;
break;
case KERN_PROC_RUID:
if (kp->ki_ruid != (uid_t)arg)
continue;
break;
}
/*
* We're going to add another proc to the set. If this
* will overflow the buffer, assume the reason is because
* nprocs (or the proc list) is corrupt and declare an error.
*/
if (cnt >= maxcnt) {
_kvm_err(kd, kd->program, "nprocs corrupt");
return (-1);
}
/*
* gather kinfo_proc
*/
kp->ki_paddr = p;
kp->ki_addr = 0; /* XXX uarea */
/* kp->ki_kstack = proc.p_thread.td_kstack; XXXKSE */
kp->ki_args = proc.p_args;
kp->ki_tracep = proc.p_tracevp;
kp->ki_textvp = proc.p_textvp;
kp->ki_fd = proc.p_fd;
kp->ki_vmspace = proc.p_vmspace;
if (proc.p_sigacts != NULL) {
if (KREAD(kd, (u_long)proc.p_sigacts, &sigacts)) {
_kvm_err(kd, kd->program,
"can't read sigacts at %p", proc.p_sigacts);
return (-1);
}
kp->ki_sigignore = sigacts.ps_sigignore;
kp->ki_sigcatch = sigacts.ps_sigcatch;
}
#if 0
if ((proc.p_flag & P_INMEM) && proc.p_stats != NULL) {
if (KREAD(kd, (u_long)proc.p_stats, &pstats)) {
_kvm_err(kd, kd->program,
"can't read stats at %x", proc.p_stats);
return (-1);
}
kp->ki_start = pstats.p_start;
/*
* XXX: The times here are probably zero and need
* to be calculated from the raw data in p_rux and
* p_crux.
*/
kp->ki_rusage = pstats.p_ru;
kp->ki_childstime = pstats.p_cru.ru_stime;
kp->ki_childutime = pstats.p_cru.ru_utime;
/* Some callers want child-times in a single value */
timeradd(&kp->ki_childstime, &kp->ki_childutime,
&kp->ki_childtime);
}
#endif
if (proc.p_oppid)
kp->ki_ppid = proc.p_oppid;
else if (proc.p_pptr) {
if (KREAD(kd, (u_long)proc.p_pptr, &pproc)) {
_kvm_err(kd, kd->program,
"can't read pproc at %p", proc.p_pptr);
return (-1);
}
kp->ki_ppid = pproc.p_pid;
} else
kp->ki_ppid = 0;
if (proc.p_pgrp == NULL)
goto nopgrp;
if (KREAD(kd, (u_long)proc.p_pgrp, &pgrp)) {
_kvm_err(kd, kd->program, "can't read pgrp at %p",
proc.p_pgrp);
return (-1);
}
kp->ki_pgid = pgrp.pg_id;
kp->ki_jobc = pgrp.pg_jobc;
if (KREAD(kd, (u_long)pgrp.pg_session, &sess)) {
_kvm_err(kd, kd->program, "can't read session at %p",
pgrp.pg_session);
return (-1);
}
kp->ki_sid = sess.s_sid;
(void)memcpy(kp->ki_login, sess.s_login,
sizeof(kp->ki_login));
kp->ki_kiflag = sess.s_ttyvp ? KI_CTTY : 0;
if (sess.s_leader == p)
kp->ki_kiflag |= KI_SLEADER;
if ((proc.p_flag & P_CONTROLT) && sess.s_ttyp != NULL) {
if (KREAD(kd, (u_long)sess.s_ttyp, &tty)) {
_kvm_err(kd, kd->program,
"can't read tty at %p", sess.s_ttyp);
return (-1);
}
if (tty.t_dev != NULL) {
if (KREAD(kd, (u_long)tty.t_dev, &t_cdev)) {
_kvm_err(kd, kd->program,
"can't read cdev at %p",
tty.t_dev);
return (-1);
}
#if 0
kp->ki_tdev = t_cdev.si_udev;
#else
kp->ki_tdev = NODEV;
#endif
}
if (tty.t_pgrp != NULL) {
if (KREAD(kd, (u_long)tty.t_pgrp, &pgrp)) {
_kvm_err(kd, kd->program,
"can't read tpgrp at %p",
tty.t_pgrp);
return (-1);
}
kp->ki_tpgid = pgrp.pg_id;
} else
kp->ki_tpgid = -1;
if (tty.t_session != NULL) {
if (KREAD(kd, (u_long)tty.t_session, &sess)) {
_kvm_err(kd, kd->program,
"can't read session at %p",
tty.t_session);
return (-1);
}
kp->ki_tsid = sess.s_sid;
}
} else {
nopgrp:
kp->ki_tdev = NODEV;
}
if ((proc.p_state != PRS_ZOMBIE) && mtd.td_wmesg)
(void)kvm_read(kd, (u_long)mtd.td_wmesg,
kp->ki_wmesg, WMESGLEN);
(void)kvm_read(kd, (u_long)proc.p_vmspace,
(char *)&vmspace, sizeof(vmspace));
kp->ki_size = vmspace.vm_map.size;
/*
* Approximate the kernel's method of calculating
* this field.
*/
#define pmap_resident_count(pm) ((pm)->pm_stats.resident_count)
kp->ki_rssize = pmap_resident_count(&vmspace.vm_pmap);
kp->ki_swrss = vmspace.vm_swrss;
kp->ki_tsize = vmspace.vm_tsize;
kp->ki_dsize = vmspace.vm_dsize;
kp->ki_ssize = vmspace.vm_ssize;
switch (what & ~KERN_PROC_INC_THREAD) {
case KERN_PROC_PGRP:
if (kp->ki_pgid != (pid_t)arg)
continue;
break;
case KERN_PROC_SESSION:
if (kp->ki_sid != (pid_t)arg)
continue;
break;
case KERN_PROC_TTY:
if ((proc.p_flag & P_CONTROLT) == 0 ||
kp->ki_tdev != (dev_t)arg)
continue;
break;
}
if (proc.p_comm[0] != 0)
strlcpy(kp->ki_comm, proc.p_comm, MAXCOMLEN);
(void)kvm_read(kd, (u_long)proc.p_sysent, (char *)&sysent,
sizeof(sysent));
(void)kvm_read(kd, (u_long)sysent.sv_name, (char *)&svname,
sizeof(svname));
if (svname[0] != 0)
strlcpy(kp->ki_emul, svname, KI_EMULNAMELEN);
if ((proc.p_state != PRS_ZOMBIE) &&
(mtd.td_blocked != 0)) {
kp->ki_kiflag |= KI_LOCKBLOCK;
if (mtd.td_lockname)
(void)kvm_read(kd,
(u_long)mtd.td_lockname,
kp->ki_lockname, LOCKNAMELEN);
kp->ki_lockname[LOCKNAMELEN] = 0;
}
kp->ki_runtime = cputick2usec(proc.p_rux.rux_runtime);
kp->ki_pid = proc.p_pid;
kp->ki_siglist = proc.p_siglist;
SIGSETOR(kp->ki_siglist, mtd.td_siglist);
kp->ki_sigmask = mtd.td_sigmask;
kp->ki_xstat = KW_EXITCODE(proc.p_xexit, proc.p_xsig);
kp->ki_acflag = proc.p_acflag;
kp->ki_lock = proc.p_lock;
if (proc.p_state != PRS_ZOMBIE) {
kp->ki_swtime = (ticks - proc.p_swtick) / hz;
kp->ki_flag = proc.p_flag;
kp->ki_sflag = 0;
kp->ki_nice = proc.p_nice;
kp->ki_traceflag = proc.p_traceflag;
if (proc.p_state == PRS_NORMAL) {
if (TD_ON_RUNQ(&mtd) ||
TD_CAN_RUN(&mtd) ||
TD_IS_RUNNING(&mtd)) {
kp->ki_stat = SRUN;
} else if (mtd.td_state ==
TDS_INHIBITED) {
if (P_SHOULDSTOP(&proc)) {
kp->ki_stat = SSTOP;
} else if (
TD_IS_SLEEPING(&mtd)) {
kp->ki_stat = SSLEEP;
} else if (TD_ON_LOCK(&mtd)) {
kp->ki_stat = SLOCK;
} else {
kp->ki_stat = SWAIT;
}
}
} else {
kp->ki_stat = SIDL;
}
/* Stuff from the thread */
kp->ki_pri.pri_level = mtd.td_priority;
kp->ki_pri.pri_native = mtd.td_base_pri;
kp->ki_lastcpu = mtd.td_lastcpu;
kp->ki_wchan = mtd.td_wchan;
kp->ki_oncpu = mtd.td_oncpu;
if (mtd.td_name[0] != '\0')
strlcpy(kp->ki_tdname, mtd.td_name, sizeof(kp->ki_tdname));
kp->ki_pctcpu = 0;
kp->ki_rqindex = 0;
/*
* Note: legacy fields; wraps at NO_CPU_OLD or the
* old max CPU value as appropriate
*/
if (mtd.td_lastcpu == NOCPU)
kp->ki_lastcpu_old = NOCPU_OLD;
else if (mtd.td_lastcpu > MAXCPU_OLD)
kp->ki_lastcpu_old = MAXCPU_OLD;
else
kp->ki_lastcpu_old = mtd.td_lastcpu;
if (mtd.td_oncpu == NOCPU)
kp->ki_oncpu_old = NOCPU_OLD;
else if (mtd.td_oncpu > MAXCPU_OLD)
kp->ki_oncpu_old = MAXCPU_OLD;
else
kp->ki_oncpu_old = mtd.td_oncpu;
} else {
kp->ki_stat = SZOMB;
}
kp->ki_tdev_freebsd11 = kp->ki_tdev; /* truncate */
bcopy(&kinfo_proc, bp, sizeof(kinfo_proc));
++bp;
++cnt;
}
return (cnt);
}