Index: head/usr.bin/top/machine.c =================================================================== --- head/usr.bin/top/machine.c (revision 334473) +++ head/usr.bin/top/machine.c (revision 334474) @@ -1,1665 +1,1665 @@ /* * top - a top users display for Unix * * DESCRIPTION: * Originally written for BSD4.4 system by Christos Zoulas. * Ported to FreeBSD 2.x by Steven Wallace && Wolfram Schneider * Order support hacked in from top-3.5beta6/machine/m_aix41.c * by Monte Mitzelfelt (for latest top see http://www.groupsys.com/topinfo/) * * AUTHOR: Christos Zoulas * Steven Wallace * Wolfram Schneider * Thomas Moestl * * $FreeBSD$ */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "top.h" #include "display.h" #include "machine.h" #include "loadavg.h" #include "screen.h" #include "utils.h" #include "layout.h" #define GETSYSCTL(name, var) getsysctl(name, &(var), sizeof(var)) #define SMPUNAMELEN 13 #define UPUNAMELEN 15 extern struct timeval timeout; static int smpmode; enum displaymodes displaymode; static int namelength = 8; /* TOP_JID_LEN based on max of 999999 */ #define TOP_JID_LEN 7 #define TOP_SWAP_LEN 6 static int jidlength; static int swaplength; static int cmdlengthdelta; /* get_process_info passes back a handle. This is what it looks like: */ struct handle { struct kinfo_proc **next_proc; /* points to next valid proc pointer */ int remaining; /* number of pointers remaining */ }; /* define what weighted cpu is. */ #define weighted_cpu(pct, pp) ((pp)->ki_swtime == 0 ? 0.0 : \ ((pct) / (1.0 - exp((pp)->ki_swtime * logcpu)))) /* what we consider to be process size: */ #define PROCSIZE(pp) ((pp)->ki_size / 1024) #define RU(pp) (&(pp)->ki_rusage) #define RUTOT(pp) \ (RU(pp)->ru_inblock + RU(pp)->ru_oublock + RU(pp)->ru_majflt) #define PCTCPU(pp) (pcpu[pp - pbase]) /* definitions for indices in the nlist array */ /* * These definitions control the format of the per-process area */ static char io_header[] = " PID%*s %-*.*s VCSW IVCSW READ WRITE FAULT TOTAL PERCENT COMMAND"; #define io_Proc_format \ "%5d%*s %-*.*s %6ld %6ld %6ld %6ld %6ld %6ld %6.2f%% %.*s" -static char smp_header_thr[] = +static char smp_header_thr_and_pid[] = " PID%*s %-*.*s THR PRI NICE SIZE RES%*s STATE C TIME %7s COMMAND"; -static char smp_header[] = - " PID%*s %-*.*s " "PRI NICE SIZE RES%*s STATE C TIME %7s COMMAND"; +static char smp_header_tid_only[] = + " THR%*s %-*.*s " "PRI NICE SIZE RES%*s STATE C TIME %7s COMMAND"; #define smp_Proc_format \ "%5d%*s %-*.*s %s%3d %4s%7s %6s%*.*s %-6.6s %2d%7s %6.2f%% %.*s" -static char up_header_thr[] = +static char up_header_thr_and_pid[] = " PID%*s %-*.*s THR PRI NICE SIZE RES%*s STATE TIME %7s COMMAND"; -static char up_header[] = - " PID%*s %-*.*s " "PRI NICE SIZE RES%*s STATE TIME %7s COMMAND"; +static char up_header_tid_only[] = + " THR%*s %-*.*s " "PRI NICE SIZE RES%*s STATE TIME %7s COMMAND"; #define up_Proc_format \ "%5d%*s %-*.*s %s%3d %4s%7s %6s%*.*s %-6.6s%.0d%7s %6.2f%% %.*s" /* process state names for the "STATE" column of the display */ /* the extra nulls in the string "run" are for adding a slash and the processor number when needed */ static char *state_abbrev[] = { "", "START", "RUN\0\0\0", "SLEEP", "STOP", "ZOMB", "WAIT", "LOCK" }; static kvm_t *kd; /* values that we stash away in _init and use in later routines */ static double logcpu; /* these are retrieved from the kernel in _init */ static load_avg ccpu; /* these are used in the get_ functions */ static int lastpid; /* these are for calculating cpu state percentages */ static long cp_time[CPUSTATES]; static long cp_old[CPUSTATES]; static long cp_diff[CPUSTATES]; /* these are for detailing the process states */ static int process_states[8]; static char *procstatenames[] = { "", " starting, ", " running, ", " sleeping, ", " stopped, ", " zombie, ", " waiting, ", " lock, ", NULL }; /* these are for detailing the cpu states */ static int cpu_states[CPUSTATES]; static char *cpustatenames[] = { "user", "nice", "system", "interrupt", "idle", NULL }; /* these are for detailing the memory statistics */ static int memory_stats[7]; static char *memorynames[] = { "K Active, ", "K Inact, ", "K Laundry, ", "K Wired, ", "K Buf, ", "K Free", NULL }; static int arc_stats[7]; static char *arcnames[] = { "K Total, ", "K MFU, ", "K MRU, ", "K Anon, ", "K Header, ", "K Other", NULL }; static int carc_stats[4]; static char *carcnames[] = { "K Compressed, ", "K Uncompressed, ", ":1 Ratio, ", NULL }; static int swap_stats[7]; static char *swapnames[] = { "K Total, ", "K Used, ", "K Free, ", "% Inuse, ", "K In, ", "K Out", NULL }; /* these are for keeping track of the proc array */ static int nproc; static int onproc = -1; static int pref_len; static struct kinfo_proc *pbase; static struct kinfo_proc **pref; static struct kinfo_proc *previous_procs; static struct kinfo_proc **previous_pref; static int previous_proc_count = 0; static int previous_proc_count_max = 0; static int previous_thread; /* data used for recalculating pctcpu */ static double *pcpu; static struct timespec proc_uptime; static struct timeval proc_wall_time; static struct timeval previous_wall_time; static uint64_t previous_interval = 0; /* total number of io operations */ static long total_inblock; static long total_oublock; static long total_majflt; /* these are for getting the memory statistics */ static int arc_enabled; static int carc_enabled; static int pageshift; /* log base 2 of the pagesize */ /* define pagetok in terms of pageshift */ #define pagetok(size) ((size) << pageshift) /* swap usage */ #define ki_swap(kip) \ ((kip)->ki_swrss > (kip)->ki_rssize ? (kip)->ki_swrss - (kip)->ki_rssize : 0) /* * Sorting orders. The first element is the default. */ char *ordernames[] = { "cpu", "size", "res", "time", "pri", "threads", "total", "read", "write", "fault", "vcsw", "ivcsw", "jid", "swap", "pid", NULL }; /* Per-cpu time states */ static int maxcpu; static int maxid; static int ncpus; static u_long cpumask; static long *times; static long *pcpu_cp_time; static long *pcpu_cp_old; static long *pcpu_cp_diff; static int *pcpu_cpu_states; static int compare_swap(const void *a, const void *b); static int compare_jid(const void *a, const void *b); static int compare_pid(const void *a, const void *b); static int compare_tid(const void *a, const void *b); static const char *format_nice(const struct kinfo_proc *pp); static void getsysctl(const char *name, void *ptr, size_t len); static int swapmode(int *retavail, int *retfree); static void update_layout(void); static int find_uid(uid_t needle, int *haystack); static int find_uid(uid_t needle, int *haystack) { size_t i = 0; for (; i < TOP_MAX_UIDS; ++i) if ((uid_t)haystack[i] == needle) return 1; return 0; } void toggle_pcpustats(void) { if (ncpus == 1) return; update_layout(); } /* Adjust display based on ncpus and the ARC state. */ static void update_layout(void) { y_mem = 3; y_arc = 4; y_carc = 5; y_swap = 4 + arc_enabled + carc_enabled; y_idlecursor = 5 + arc_enabled + carc_enabled; y_message = 5 + arc_enabled + carc_enabled; y_header = 6 + arc_enabled + carc_enabled; y_procs = 7 + arc_enabled + carc_enabled; Header_lines = 7 + arc_enabled + carc_enabled; if (pcpu_stats) { y_mem += ncpus - 1; y_arc += ncpus - 1; y_carc += ncpus - 1; y_swap += ncpus - 1; y_idlecursor += ncpus - 1; y_message += ncpus - 1; y_header += ncpus - 1; y_procs += ncpus - 1; Header_lines += ncpus - 1; } } int machine_init(struct statics *statics) { int i, j, empty, pagesize; uint64_t arc_size; boolean_t carc_en; size_t size; size = sizeof(smpmode); if ((sysctlbyname("machdep.smp_active", &smpmode, &size, NULL, 0) != 0 && sysctlbyname("kern.smp.active", &smpmode, &size, NULL, 0) != 0) || size != sizeof(smpmode)) smpmode = 0; size = sizeof(arc_size); if (sysctlbyname("kstat.zfs.misc.arcstats.size", &arc_size, &size, NULL, 0) == 0 && arc_size != 0) arc_enabled = 1; size = sizeof(carc_en); if (arc_enabled && sysctlbyname("vfs.zfs.compressed_arc_enabled", &carc_en, &size, NULL, 0) == 0 && carc_en == 1) carc_enabled = 1; namelength = MAXLOGNAME; if (smpmode && namelength > SMPUNAMELEN) namelength = SMPUNAMELEN; else if (namelength > UPUNAMELEN) namelength = UPUNAMELEN; kd = kvm_open(NULL, _PATH_DEVNULL, NULL, O_RDONLY, "kvm_open"); if (kd == NULL) return (-1); GETSYSCTL("kern.ccpu", ccpu); /* this is used in calculating WCPU -- calculate it ahead of time */ logcpu = log(loaddouble(ccpu)); pbase = NULL; pref = NULL; pcpu = NULL; nproc = 0; onproc = -1; /* get the page size and calculate pageshift from it */ pagesize = getpagesize(); pageshift = 0; while (pagesize > 1) { pageshift++; pagesize >>= 1; } /* we only need the amount of log(2)1024 for our conversion */ pageshift -= LOG1024; /* fill in the statics information */ statics->procstate_names = procstatenames; statics->cpustate_names = cpustatenames; statics->memory_names = memorynames; if (arc_enabled) statics->arc_names = arcnames; else statics->arc_names = NULL; if (carc_enabled) statics->carc_names = carcnames; else statics->carc_names = NULL; statics->swap_names = swapnames; statics->order_names = ordernames; /* Allocate state for per-CPU stats. */ cpumask = 0; ncpus = 0; GETSYSCTL("kern.smp.maxcpus", maxcpu); size = sizeof(long) * maxcpu * CPUSTATES; times = malloc(size); if (times == NULL) err(1, "malloc %zu bytes", size); if (sysctlbyname("kern.cp_times", times, &size, NULL, 0) == -1) err(1, "sysctlbyname kern.cp_times"); pcpu_cp_time = calloc(1, size); maxid = (size / CPUSTATES / sizeof(long)) - 1; for (i = 0; i <= maxid; i++) { empty = 1; for (j = 0; empty && j < CPUSTATES; j++) { if (times[i * CPUSTATES + j] != 0) empty = 0; } if (!empty) { cpumask |= (1ul << i); ncpus++; } } size = sizeof(long) * ncpus * CPUSTATES; pcpu_cp_old = calloc(1, size); pcpu_cp_diff = calloc(1, size); pcpu_cpu_states = calloc(1, size); statics->ncpus = ncpus; update_layout(); /* all done! */ return (0); } char * format_header(char *uname_field) { static char Header[128]; const char *prehead; if (ps.jail) jidlength = TOP_JID_LEN + 1; /* +1 for extra left space. */ else jidlength = 0; if (ps.swap) swaplength = TOP_SWAP_LEN + 1; /* +1 for extra left space */ else swaplength = 0; switch (displaymode) { case DISP_CPU: /* * The logic of picking the right header format seems reverse * here because we only want to display a THR column when * "thread mode" is off (and threads are not listed as * separate lines). */ prehead = smpmode ? - (ps.thread ? smp_header : smp_header_thr) : - (ps.thread ? up_header : up_header_thr); + (ps.thread ? smp_header_tid_only : smp_header_thr_and_pid) : + (ps.thread ? up_header_tid_only : up_header_thr_and_pid); snprintf(Header, sizeof(Header), prehead, jidlength, ps.jail ? " JID" : "", namelength, namelength, uname_field, swaplength, ps.swap ? " SWAP" : "", ps.wcpu ? "WCPU" : "CPU"); break; case DISP_IO: prehead = io_header; snprintf(Header, sizeof(Header), prehead, jidlength, ps.jail ? " JID" : "", namelength, namelength, uname_field); break; case DISP_MAX: assert("displaymode must not be set to DISP_MAX"); } cmdlengthdelta = strlen(Header) - 7; return (Header); } static int swappgsin = -1; static int swappgsout = -1; void get_system_info(struct system_info *si) { struct loadavg sysload; int mib[2]; struct timeval boottime; uint64_t arc_stat, arc_stat2; int i, j; size_t size; /* get the CPU stats */ size = (maxid + 1) * CPUSTATES * sizeof(long); if (sysctlbyname("kern.cp_times", pcpu_cp_time, &size, NULL, 0) == -1) err(1, "sysctlbyname kern.cp_times"); GETSYSCTL("kern.cp_time", cp_time); GETSYSCTL("vm.loadavg", sysload); GETSYSCTL("kern.lastpid", lastpid); /* convert load averages to doubles */ for (i = 0; i < 3; i++) si->load_avg[i] = (double)sysload.ldavg[i] / sysload.fscale; /* convert cp_time counts to percentages */ for (i = j = 0; i <= maxid; i++) { if ((cpumask & (1ul << i)) == 0) continue; percentages(CPUSTATES, &pcpu_cpu_states[j * CPUSTATES], &pcpu_cp_time[j * CPUSTATES], &pcpu_cp_old[j * CPUSTATES], &pcpu_cp_diff[j * CPUSTATES]); j++; } percentages(CPUSTATES, cpu_states, cp_time, cp_old, cp_diff); /* sum memory & swap statistics */ { static unsigned int swap_delay = 0; static int swapavail = 0; static int swapfree = 0; static long bufspace = 0; static uint64_t nspgsin, nspgsout; GETSYSCTL("vfs.bufspace", bufspace); GETSYSCTL("vm.stats.vm.v_active_count", memory_stats[0]); GETSYSCTL("vm.stats.vm.v_inactive_count", memory_stats[1]); GETSYSCTL("vm.stats.vm.v_laundry_count", memory_stats[2]); GETSYSCTL("vm.stats.vm.v_wire_count", memory_stats[3]); GETSYSCTL("vm.stats.vm.v_free_count", memory_stats[5]); GETSYSCTL("vm.stats.vm.v_swappgsin", nspgsin); GETSYSCTL("vm.stats.vm.v_swappgsout", nspgsout); /* convert memory stats to Kbytes */ memory_stats[0] = pagetok(memory_stats[0]); memory_stats[1] = pagetok(memory_stats[1]); memory_stats[2] = pagetok(memory_stats[2]); memory_stats[3] = pagetok(memory_stats[3]); memory_stats[4] = bufspace / 1024; memory_stats[5] = pagetok(memory_stats[5]); memory_stats[6] = -1; /* first interval */ if (swappgsin < 0) { swap_stats[4] = 0; swap_stats[5] = 0; } /* compute differences between old and new swap statistic */ else { swap_stats[4] = pagetok(((nspgsin - swappgsin))); swap_stats[5] = pagetok(((nspgsout - swappgsout))); } swappgsin = nspgsin; swappgsout = nspgsout; /* call CPU heavy swapmode() only for changes */ if (swap_stats[4] > 0 || swap_stats[5] > 0 || swap_delay == 0) { swap_stats[3] = swapmode(&swapavail, &swapfree); swap_stats[0] = swapavail; swap_stats[1] = swapavail - swapfree; swap_stats[2] = swapfree; } swap_delay = 1; swap_stats[6] = -1; } if (arc_enabled) { GETSYSCTL("kstat.zfs.misc.arcstats.size", arc_stat); arc_stats[0] = arc_stat >> 10; GETSYSCTL("vfs.zfs.mfu_size", arc_stat); arc_stats[1] = arc_stat >> 10; GETSYSCTL("vfs.zfs.mru_size", arc_stat); arc_stats[2] = arc_stat >> 10; GETSYSCTL("vfs.zfs.anon_size", arc_stat); arc_stats[3] = arc_stat >> 10; GETSYSCTL("kstat.zfs.misc.arcstats.hdr_size", arc_stat); GETSYSCTL("kstat.zfs.misc.arcstats.l2_hdr_size", arc_stat2); arc_stats[4] = (arc_stat + arc_stat2) >> 10; GETSYSCTL("kstat.zfs.misc.arcstats.other_size", arc_stat); arc_stats[5] = arc_stat >> 10; si->arc = arc_stats; } if (carc_enabled) { GETSYSCTL("kstat.zfs.misc.arcstats.compressed_size", arc_stat); carc_stats[0] = arc_stat >> 10; carc_stats[2] = arc_stat >> 10; /* For ratio */ GETSYSCTL("kstat.zfs.misc.arcstats.uncompressed_size", arc_stat); carc_stats[1] = arc_stat >> 10; si->carc = carc_stats; } /* set arrays and strings */ if (pcpu_stats) { si->cpustates = pcpu_cpu_states; si->ncpus = ncpus; } else { si->cpustates = cpu_states; si->ncpus = 1; } si->memory = memory_stats; si->swap = swap_stats; if (lastpid > 0) { si->last_pid = lastpid; } else { si->last_pid = -1; } /* * Print how long system has been up. * (Found by looking getting "boottime" from the kernel) */ mib[0] = CTL_KERN; mib[1] = KERN_BOOTTIME; size = sizeof(boottime); if (sysctl(mib, nitems(mib), &boottime, &size, NULL, 0) != -1 && boottime.tv_sec != 0) { si->boottime = boottime; } else { si->boottime.tv_sec = -1; } } #define NOPROC ((void *)-1) /* * We need to compare data from the old process entry with the new * process entry. * To facilitate doing this quickly we stash a pointer in the kinfo_proc * structure to cache the mapping. We also use a negative cache pointer * of NOPROC to avoid duplicate lookups. * XXX: this could be done when the actual processes are fetched, we do * it here out of laziness. */ static const struct kinfo_proc * get_old_proc(struct kinfo_proc *pp) { struct kinfo_proc **oldpp, *oldp; /* * If this is the first fetch of the kinfo_procs then we don't have * any previous entries. */ if (previous_proc_count == 0) return (NULL); /* negative cache? */ if (pp->ki_udata == NOPROC) return (NULL); /* cached? */ if (pp->ki_udata != NULL) return (pp->ki_udata); /* * Not cached, * 1) look up based on pid. * 2) compare process start. * If we fail here, then setup a negative cache entry, otherwise * cache it. */ oldpp = bsearch(&pp, previous_pref, previous_proc_count, sizeof(*previous_pref), ps.thread ? compare_tid : compare_pid); if (oldpp == NULL) { pp->ki_udata = NOPROC; return (NULL); } oldp = *oldpp; if (bcmp(&oldp->ki_start, &pp->ki_start, sizeof(pp->ki_start)) != 0) { pp->ki_udata = NOPROC; return (NULL); } pp->ki_udata = oldp; return (oldp); } /* * Return the total amount of IO done in blocks in/out and faults. * store the values individually in the pointers passed in. */ static long get_io_stats(struct kinfo_proc *pp, long *inp, long *oup, long *flp, long *vcsw, long *ivcsw) { const struct kinfo_proc *oldp; static struct kinfo_proc dummy; long ret; oldp = get_old_proc(pp); if (oldp == NULL) { bzero(&dummy, sizeof(dummy)); oldp = &dummy; } *inp = RU(pp)->ru_inblock - RU(oldp)->ru_inblock; *oup = RU(pp)->ru_oublock - RU(oldp)->ru_oublock; *flp = RU(pp)->ru_majflt - RU(oldp)->ru_majflt; *vcsw = RU(pp)->ru_nvcsw - RU(oldp)->ru_nvcsw; *ivcsw = RU(pp)->ru_nivcsw - RU(oldp)->ru_nivcsw; ret = (RU(pp)->ru_inblock - RU(oldp)->ru_inblock) + (RU(pp)->ru_oublock - RU(oldp)->ru_oublock) + (RU(pp)->ru_majflt - RU(oldp)->ru_majflt); return (ret); } /* * If there was a previous update, use the delta in ki_runtime over * the previous interval to calculate pctcpu. Otherwise, fall back * to using the kernel's ki_pctcpu. */ static double proc_calc_pctcpu(struct kinfo_proc *pp) { const struct kinfo_proc *oldp; if (previous_interval != 0) { oldp = get_old_proc(pp); if (oldp != NULL) return ((double)(pp->ki_runtime - oldp->ki_runtime) / previous_interval); /* * If this process/thread was created during the previous * interval, charge it's total runtime to the previous * interval. */ else if (pp->ki_start.tv_sec > previous_wall_time.tv_sec || (pp->ki_start.tv_sec == previous_wall_time.tv_sec && pp->ki_start.tv_usec >= previous_wall_time.tv_usec)) return ((double)pp->ki_runtime / previous_interval); } return (pctdouble(pp->ki_pctcpu)); } /* * Return true if this process has used any CPU time since the * previous update. */ static int proc_used_cpu(struct kinfo_proc *pp) { const struct kinfo_proc *oldp; oldp = get_old_proc(pp); if (oldp == NULL) return (PCTCPU(pp) != 0); return (pp->ki_runtime != oldp->ki_runtime || RU(pp)->ru_nvcsw != RU(oldp)->ru_nvcsw || RU(pp)->ru_nivcsw != RU(oldp)->ru_nivcsw); } /* * Return the total number of block in/out and faults by a process. */ static long get_io_total(struct kinfo_proc *pp) { long dummy; return (get_io_stats(pp, &dummy, &dummy, &dummy, &dummy, &dummy)); } static struct handle handle; void * get_process_info(struct system_info *si, struct process_select *sel, int (*compare)(const void *, const void *)) { int i; int total_procs; long p_io; long p_inblock, p_oublock, p_majflt, p_vcsw, p_ivcsw; long nsec; int active_procs; struct kinfo_proc **prefp; struct kinfo_proc *pp; struct timespec previous_proc_uptime; /* these are copied out of sel for speed */ int show_idle; int show_jid; int show_self; int show_system; int show_uid; int show_kidle; /* * If thread state was toggled, don't cache the previous processes. */ if (previous_thread != sel->thread) nproc = 0; previous_thread = sel->thread; /* * Save the previous process info. */ if (previous_proc_count_max < nproc) { free(previous_procs); previous_procs = malloc(nproc * sizeof(*previous_procs)); free(previous_pref); previous_pref = malloc(nproc * sizeof(*previous_pref)); if (previous_procs == NULL || previous_pref == NULL) { (void) fprintf(stderr, "top: Out of memory.\n"); quit(23); } previous_proc_count_max = nproc; } if (nproc) { for (i = 0; i < nproc; i++) previous_pref[i] = &previous_procs[i]; bcopy(pbase, previous_procs, nproc * sizeof(*previous_procs)); qsort(previous_pref, nproc, sizeof(*previous_pref), ps.thread ? compare_tid : compare_pid); } previous_proc_count = nproc; previous_proc_uptime = proc_uptime; previous_wall_time = proc_wall_time; previous_interval = 0; pbase = kvm_getprocs(kd, sel->thread ? KERN_PROC_ALL : KERN_PROC_PROC, 0, &nproc); (void)gettimeofday(&proc_wall_time, NULL); if (clock_gettime(CLOCK_UPTIME, &proc_uptime) != 0) memset(&proc_uptime, 0, sizeof(proc_uptime)); else if (previous_proc_uptime.tv_sec != 0 && previous_proc_uptime.tv_nsec != 0) { previous_interval = (proc_uptime.tv_sec - previous_proc_uptime.tv_sec) * 1000000; nsec = proc_uptime.tv_nsec - previous_proc_uptime.tv_nsec; if (nsec < 0) { previous_interval -= 1000000; nsec += 1000000000; } previous_interval += nsec / 1000; } if (nproc > onproc) { pref = realloc(pref, sizeof(*pref) * nproc); pcpu = realloc(pcpu, sizeof(*pcpu) * nproc); onproc = nproc; } if (pref == NULL || pbase == NULL || pcpu == NULL) { (void) fprintf(stderr, "top: Out of memory.\n"); quit(23); } /* get a pointer to the states summary array */ si->procstates = process_states; /* set up flags which define what we are going to select */ show_idle = sel->idle; show_jid = sel->jid != -1; show_self = sel->self == -1; show_system = sel->system; show_uid = sel->uid[0] != -1; show_kidle = sel->kidle; /* count up process states and get pointers to interesting procs */ total_procs = 0; active_procs = 0; total_inblock = 0; total_oublock = 0; total_majflt = 0; memset((char *)process_states, 0, sizeof(process_states)); prefp = pref; for (pp = pbase, i = 0; i < nproc; pp++, i++) { if (pp->ki_stat == 0) /* not in use */ continue; if (!show_self && pp->ki_pid == sel->self) /* skip self */ continue; if (!show_system && (pp->ki_flag & P_SYSTEM)) /* skip system process */ continue; p_io = get_io_stats(pp, &p_inblock, &p_oublock, &p_majflt, &p_vcsw, &p_ivcsw); total_inblock += p_inblock; total_oublock += p_oublock; total_majflt += p_majflt; total_procs++; process_states[(unsigned char)pp->ki_stat]++; if (pp->ki_stat == SZOMB) /* skip zombies */ continue; if (!show_kidle && pp->ki_tdflags & TDF_IDLETD) /* skip kernel idle process */ continue; PCTCPU(pp) = proc_calc_pctcpu(pp); if (sel->thread && PCTCPU(pp) > 1.0) PCTCPU(pp) = 1.0; if (displaymode == DISP_CPU && !show_idle && (!proc_used_cpu(pp) || pp->ki_stat == SSTOP || pp->ki_stat == SIDL)) /* skip idle or non-running processes */ continue; if (displaymode == DISP_IO && !show_idle && p_io == 0) /* skip processes that aren't doing I/O */ continue; if (show_jid && pp->ki_jid != sel->jid) /* skip proc. that don't belong to the selected JID */ continue; if (show_uid && !find_uid(pp->ki_ruid, sel->uid)) /* skip proc. that don't belong to the selected UID */ continue; *prefp++ = pp; active_procs++; } /* if requested, sort the "interesting" processes */ if (compare != NULL) qsort(pref, active_procs, sizeof(*pref), compare); /* remember active and total counts */ si->p_total = total_procs; si->p_pactive = pref_len = active_procs; /* pass back a handle */ handle.next_proc = pref; handle.remaining = active_procs; return ((caddr_t)&handle); } static char fmt[512]; /* static area where result is built */ char * format_next_process(caddr_t xhandle, char *(*get_userid)(int), int flags) { struct kinfo_proc *pp; const struct kinfo_proc *oldp; long cputime; double pct; struct handle *hp; char status[22]; int cpu; size_t state; struct rusage ru, *rup; long p_tot, s_tot; char *proc_fmt; char thr_buf[6]; char jid_buf[TOP_JID_LEN + 1], swap_buf[TOP_SWAP_LEN + 1]; char *cmdbuf = NULL; char **args; const int cmdlen = 128; /* find and remember the next proc structure */ hp = (struct handle *)xhandle; pp = *(hp->next_proc++); hp->remaining--; /* get the process's command name */ if ((pp->ki_flag & P_INMEM) == 0) { /* * Print swapped processes as */ size_t len; len = strlen(pp->ki_comm); if (len > sizeof(pp->ki_comm) - 3) len = sizeof(pp->ki_comm) - 3; memmove(pp->ki_comm + 1, pp->ki_comm, len); pp->ki_comm[0] = '<'; pp->ki_comm[len + 1] = '>'; pp->ki_comm[len + 2] = '\0'; } /* * Convert the process's runtime from microseconds to seconds. This * time includes the interrupt time although that is not wanted here. * ps(1) is similarly sloppy. */ cputime = (pp->ki_runtime + 500000) / 1000000; /* calculate the base for cpu percentages */ pct = PCTCPU(pp); /* generate "STATE" field */ switch (state = pp->ki_stat) { case SRUN: if (smpmode && pp->ki_oncpu != NOCPU) sprintf(status, "CPU%d", pp->ki_oncpu); else strcpy(status, "RUN"); break; case SLOCK: if (pp->ki_kiflag & KI_LOCKBLOCK) { sprintf(status, "*%.6s", pp->ki_lockname); break; } /* fall through */ case SSLEEP: sprintf(status, "%.6s", pp->ki_wmesg); break; default: if (state < sizeof(state_abbrev) / sizeof(*state_abbrev)) sprintf(status, "%.6s", state_abbrev[state]); else sprintf(status, "?%5zu", state); break; } cmdbuf = (char *)malloc(cmdlen + 1); if (cmdbuf == NULL) { warn("malloc(%d)", cmdlen + 1); return NULL; } if (!(flags & FMT_SHOWARGS)) { if (ps.thread && pp->ki_flag & P_HADTHREADS && pp->ki_tdname[0]) { snprintf(cmdbuf, cmdlen, "%s{%s%s}", pp->ki_comm, pp->ki_tdname, pp->ki_moretdname); } else { snprintf(cmdbuf, cmdlen, "%s", pp->ki_comm); } } else { if (pp->ki_flag & P_SYSTEM || pp->ki_args == NULL || (args = kvm_getargv(kd, pp, cmdlen)) == NULL || !(*args)) { if (ps.thread && pp->ki_flag & P_HADTHREADS && pp->ki_tdname[0]) { snprintf(cmdbuf, cmdlen, "[%s{%s%s}]", pp->ki_comm, pp->ki_tdname, pp->ki_moretdname); } else { snprintf(cmdbuf, cmdlen, "[%s]", pp->ki_comm); } } else { char *src, *dst, *argbuf; char *cmd; size_t argbuflen; size_t len; argbuflen = cmdlen * 4; argbuf = (char *)malloc(argbuflen + 1); if (argbuf == NULL) { warn("malloc(%zu)", argbuflen + 1); free(cmdbuf); return NULL; } dst = argbuf; /* Extract cmd name from argv */ cmd = strrchr(*args, '/'); if (cmd == NULL) cmd = *args; else cmd++; for (; (src = *args++) != NULL; ) { if (*src == '\0') continue; len = (argbuflen - (dst - argbuf) - 1) / 4; strvisx(dst, src, MIN(strlen(src), len), VIS_NL | VIS_CSTYLE); while (*dst != '\0') dst++; if ((argbuflen - (dst - argbuf) - 1) / 4 > 0) *dst++ = ' '; /* add delimiting space */ } if (dst != argbuf && dst[-1] == ' ') dst--; *dst = '\0'; if (strcmp(cmd, pp->ki_comm) != 0) { if (ps.thread && pp->ki_flag & P_HADTHREADS && pp->ki_tdname[0]) snprintf(cmdbuf, cmdlen, "%s (%s){%s%s}", argbuf, pp->ki_comm, pp->ki_tdname, pp->ki_moretdname); else snprintf(cmdbuf, cmdlen, "%s (%s)", argbuf, pp->ki_comm); } else { if (ps.thread && pp->ki_flag & P_HADTHREADS && pp->ki_tdname[0]) snprintf(cmdbuf, cmdlen, "%s{%s%s}", argbuf, pp->ki_tdname, pp->ki_moretdname); else strlcpy(cmdbuf, argbuf, cmdlen); } free(argbuf); } } if (ps.jail == 0) jid_buf[0] = '\0'; else snprintf(jid_buf, sizeof(jid_buf), "%*d", jidlength - 1, pp->ki_jid); if (ps.swap == 0) swap_buf[0] = '\0'; else snprintf(swap_buf, sizeof(swap_buf), "%*s", swaplength - 1, format_k2(pagetok(ki_swap(pp)))); /* XXX */ if (displaymode == DISP_IO) { oldp = get_old_proc(pp); if (oldp != NULL) { ru.ru_inblock = RU(pp)->ru_inblock - RU(oldp)->ru_inblock; ru.ru_oublock = RU(pp)->ru_oublock - RU(oldp)->ru_oublock; ru.ru_majflt = RU(pp)->ru_majflt - RU(oldp)->ru_majflt; ru.ru_nvcsw = RU(pp)->ru_nvcsw - RU(oldp)->ru_nvcsw; ru.ru_nivcsw = RU(pp)->ru_nivcsw - RU(oldp)->ru_nivcsw; rup = &ru; } else { rup = RU(pp); } p_tot = rup->ru_inblock + rup->ru_oublock + rup->ru_majflt; s_tot = total_inblock + total_oublock + total_majflt; snprintf(fmt, sizeof(fmt), io_Proc_format, pp->ki_pid, jidlength, jid_buf, namelength, namelength, (*get_userid)(pp->ki_ruid), rup->ru_nvcsw, rup->ru_nivcsw, rup->ru_inblock, rup->ru_oublock, rup->ru_majflt, p_tot, s_tot == 0 ? 0.0 : (p_tot * 100.0 / s_tot), screen_width > cmdlengthdelta ? screen_width - cmdlengthdelta : 0, printable(cmdbuf)); free(cmdbuf); return (fmt); } /* format this entry */ if (smpmode) { if (state == SRUN && pp->ki_oncpu != NOCPU) cpu = pp->ki_oncpu; else cpu = pp->ki_lastcpu; } else cpu = 0; proc_fmt = smpmode ? smp_Proc_format : up_Proc_format; if (ps.thread != 0) thr_buf[0] = '\0'; else snprintf(thr_buf, sizeof(thr_buf), "%*d ", (int)(sizeof(thr_buf) - 2), pp->ki_numthreads); snprintf(fmt, sizeof(fmt), proc_fmt, - pp->ki_pid, + (ps.thread) ? pp->ki_tid : pp->ki_pid, jidlength, jid_buf, namelength, namelength, (*get_userid)(pp->ki_ruid), thr_buf, pp->ki_pri.pri_level - PZERO, format_nice(pp), format_k2(PROCSIZE(pp)), format_k2(pagetok(pp->ki_rssize)), swaplength, swaplength, swap_buf, status, cpu, format_time(cputime), ps.wcpu ? 100.0 * weighted_cpu(pct, pp) : 100.0 * pct, screen_width > cmdlengthdelta ? screen_width - cmdlengthdelta : 0, printable(cmdbuf)); free(cmdbuf); /* return the result */ return (fmt); } static void getsysctl(const char *name, void *ptr, size_t len) { size_t nlen = len; if (sysctlbyname(name, ptr, &nlen, NULL, 0) == -1) { fprintf(stderr, "top: sysctl(%s...) failed: %s\n", name, strerror(errno)); quit(23); } if (nlen != len) { fprintf(stderr, "top: sysctl(%s...) expected %lu, got %lu\n", name, (unsigned long)len, (unsigned long)nlen); quit(23); } } static const char * format_nice(const struct kinfo_proc *pp) { const char *fifo, *kproc; int rtpri; static char nicebuf[4 + 1]; fifo = PRI_NEED_RR(pp->ki_pri.pri_class) ? "" : "F"; kproc = (pp->ki_flag & P_KPROC) ? "k" : ""; switch (PRI_BASE(pp->ki_pri.pri_class)) { case PRI_ITHD: return ("-"); case PRI_REALTIME: /* * XXX: the kernel doesn't tell us the original rtprio and * doesn't really know what it was, so to recover it we * must be more chummy with the implementation than the * implementation is with itself. pri_user gives a * constant "base" priority, but is only initialized * properly for user threads. pri_native gives what the * kernel calls the "base" priority, but it isn't constant * since it is changed by priority propagation. pri_native * also isn't properly initialized for all threads, but it * is properly initialized for kernel realtime and idletime * threads. Thus we use pri_user for the base priority of * user threads (it is always correct) and pri_native for * the base priority of kernel realtime and idletime threads * (there is nothing better, and it is usually correct). * * The field width and thus the buffer are too small for * values like "kr31F", but such values shouldn't occur, * and if they do then the tailing "F" is not displayed. */ rtpri = ((pp->ki_flag & P_KPROC) ? pp->ki_pri.pri_native : pp->ki_pri.pri_user) - PRI_MIN_REALTIME; snprintf(nicebuf, sizeof(nicebuf), "%sr%d%s", kproc, rtpri, fifo); break; case PRI_TIMESHARE: if (pp->ki_flag & P_KPROC) return ("-"); snprintf(nicebuf, sizeof(nicebuf), "%d", pp->ki_nice - NZERO); break; case PRI_IDLE: /* XXX: as above. */ rtpri = ((pp->ki_flag & P_KPROC) ? pp->ki_pri.pri_native : pp->ki_pri.pri_user) - PRI_MIN_IDLE; snprintf(nicebuf, sizeof(nicebuf), "%si%d%s", kproc, rtpri, fifo); break; default: return ("?"); } return (nicebuf); } /* comparison routines for qsort */ static int compare_pid(const void *p1, const void *p2) { const struct kinfo_proc * const *pp1 = p1; const struct kinfo_proc * const *pp2 = p2; if ((*pp2)->ki_pid < 0 || (*pp1)->ki_pid < 0) abort(); return ((*pp1)->ki_pid - (*pp2)->ki_pid); } static int compare_tid(const void *p1, const void *p2) { const struct kinfo_proc * const *pp1 = p1; const struct kinfo_proc * const *pp2 = p2; if ((*pp2)->ki_tid < 0 || (*pp1)->ki_tid < 0) abort(); return ((*pp1)->ki_tid - (*pp2)->ki_tid); } /* * proc_compare - comparison function for "qsort" * Compares the resource consumption of two processes using five * distinct keys. The keys (in descending order of importance) are: * percent cpu, cpu ticks, state, resident set size, total virtual * memory usage. The process states are ordered as follows (from least * to most important): WAIT, zombie, sleep, stop, start, run. The * array declaration below maps a process state index into a number * that reflects this ordering. */ static int sorted_state[] = { 0, /* not used */ 3, /* sleep */ 1, /* ABANDONED (WAIT) */ 6, /* run */ 5, /* start */ 2, /* zombie */ 4 /* stop */ }; #define ORDERKEY_PCTCPU(a, b) do { \ double diff; \ if (ps.wcpu) \ diff = weighted_cpu(PCTCPU((b)), (b)) - \ weighted_cpu(PCTCPU((a)), (a)); \ else \ diff = PCTCPU((b)) - PCTCPU((a)); \ if (diff != 0) \ return (diff > 0 ? 1 : -1); \ } while (0) #define ORDERKEY_CPTICKS(a, b) do { \ int64_t diff = (int64_t)(b)->ki_runtime - (int64_t)(a)->ki_runtime; \ if (diff != 0) \ return (diff > 0 ? 1 : -1); \ } while (0) #define ORDERKEY_STATE(a, b) do { \ int diff = sorted_state[(unsigned char)(b)->ki_stat] - sorted_state[(unsigned char)(a)->ki_stat]; \ if (diff != 0) \ return (diff > 0 ? 1 : -1); \ } while (0) #define ORDERKEY_PRIO(a, b) do { \ int diff = (int)(b)->ki_pri.pri_level - (int)(a)->ki_pri.pri_level; \ if (diff != 0) \ return (diff > 0 ? 1 : -1); \ } while (0) #define ORDERKEY_THREADS(a, b) do { \ int diff = (int)(b)->ki_numthreads - (int)(a)->ki_numthreads; \ if (diff != 0) \ return (diff > 0 ? 1 : -1); \ } while (0) #define ORDERKEY_RSSIZE(a, b) do { \ long diff = (long)(b)->ki_rssize - (long)(a)->ki_rssize; \ if (diff != 0) \ return (diff > 0 ? 1 : -1); \ } while (0) #define ORDERKEY_MEM(a, b) do { \ long diff = (long)PROCSIZE((b)) - (long)PROCSIZE((a)); \ if (diff != 0) \ return (diff > 0 ? 1 : -1); \ } while (0) #define ORDERKEY_JID(a, b) do { \ int diff = (int)(b)->ki_jid - (int)(a)->ki_jid; \ if (diff != 0) \ return (diff > 0 ? 1 : -1); \ } while (0) #define ORDERKEY_SWAP(a, b) do { \ int diff = (int)ki_swap(b) - (int)ki_swap(a); \ if (diff != 0) \ return (diff > 0 ? 1 : -1); \ } while (0) /* compare_cpu - the comparison function for sorting by cpu percentage */ static int compare_cpu(const void *arg1, const void *arg2) { struct kinfo_proc *p1 = *(struct kinfo_proc **)arg1; struct kinfo_proc *p2 = *(struct kinfo_proc **)arg2; ORDERKEY_PCTCPU(p1, p2); ORDERKEY_CPTICKS(p1, p2); ORDERKEY_STATE(p1, p2); ORDERKEY_PRIO(p1, p2); ORDERKEY_RSSIZE(p1, p2); ORDERKEY_MEM(p1, p2); return (0); } /* "cpu" compare routines */ static int compare_size(const void *arg1, const void *arg2); static int compare_res(const void *arg1, const void *arg2); static int compare_time(const void *arg1, const void *arg2); static int compare_prio(const void *arg1, const void *arg2); static int compare_threads(const void *arg1, const void *arg2); /* * "io" compare routines. Context switches aren't i/o, but are displayed * on the "io" display. */ static int compare_iototal(const void *arg1, const void *arg2); static int compare_ioread(const void *arg1, const void *arg2); static int compare_iowrite(const void *arg1, const void *arg2); static int compare_iofault(const void *arg1, const void *arg2); static int compare_vcsw(const void *arg1, const void *arg2); static int compare_ivcsw(const void *arg1, const void *arg2); int (*compares[])(const void *arg1, const void *arg2) = { compare_cpu, compare_size, compare_res, compare_time, compare_prio, compare_threads, compare_iototal, compare_ioread, compare_iowrite, compare_iofault, compare_vcsw, compare_ivcsw, compare_jid, compare_swap, NULL }; /* compare_size - the comparison function for sorting by total memory usage */ int compare_size(const void *arg1, const void *arg2) { struct kinfo_proc *p1 = *(struct kinfo_proc **)arg1; struct kinfo_proc *p2 = *(struct kinfo_proc **)arg2; ORDERKEY_MEM(p1, p2); ORDERKEY_RSSIZE(p1, p2); ORDERKEY_PCTCPU(p1, p2); ORDERKEY_CPTICKS(p1, p2); ORDERKEY_STATE(p1, p2); ORDERKEY_PRIO(p1, p2); return (0); } /* compare_res - the comparison function for sorting by resident set size */ int compare_res(const void *arg1, const void *arg2) { struct kinfo_proc *p1 = *(struct kinfo_proc **)arg1; struct kinfo_proc *p2 = *(struct kinfo_proc **)arg2; ORDERKEY_RSSIZE(p1, p2); ORDERKEY_MEM(p1, p2); ORDERKEY_PCTCPU(p1, p2); ORDERKEY_CPTICKS(p1, p2); ORDERKEY_STATE(p1, p2); ORDERKEY_PRIO(p1, p2); return (0); } /* compare_time - the comparison function for sorting by total cpu time */ int compare_time(const void *arg1, const void *arg2) { struct kinfo_proc *p1 = *(struct kinfo_proc **)arg1; struct kinfo_proc *p2 = *(struct kinfo_proc **)arg2; ORDERKEY_CPTICKS(p1, p2); ORDERKEY_PCTCPU(p1, p2); ORDERKEY_STATE(p1, p2); ORDERKEY_PRIO(p1, p2); ORDERKEY_RSSIZE(p1, p2); ORDERKEY_MEM(p1, p2); return (0); } /* compare_prio - the comparison function for sorting by priority */ int compare_prio(const void *arg1, const void *arg2) { struct kinfo_proc *p1 = *(struct kinfo_proc **)arg1; struct kinfo_proc *p2 = *(struct kinfo_proc **)arg2; ORDERKEY_PRIO(p1, p2); ORDERKEY_CPTICKS(p1, p2); ORDERKEY_PCTCPU(p1, p2); ORDERKEY_STATE(p1, p2); ORDERKEY_RSSIZE(p1, p2); ORDERKEY_MEM(p1, p2); return (0); } /* compare_threads - the comparison function for sorting by threads */ static int compare_threads(const void *arg1, const void *arg2) { struct kinfo_proc *p1 = *(struct kinfo_proc **)arg1; struct kinfo_proc *p2 = *(struct kinfo_proc **)arg2; ORDERKEY_THREADS(p1, p2); ORDERKEY_PCTCPU(p1, p2); ORDERKEY_CPTICKS(p1, p2); ORDERKEY_STATE(p1, p2); ORDERKEY_PRIO(p1, p2); ORDERKEY_RSSIZE(p1, p2); ORDERKEY_MEM(p1, p2); return (0); } /* compare_jid - the comparison function for sorting by jid */ static int compare_jid(const void *arg1, const void *arg2) { struct kinfo_proc *p1 = *(struct kinfo_proc **)arg1; struct kinfo_proc *p2 = *(struct kinfo_proc **)arg2; ORDERKEY_JID(p1, p2); ORDERKEY_PCTCPU(p1, p2); ORDERKEY_CPTICKS(p1, p2); ORDERKEY_STATE(p1, p2); ORDERKEY_PRIO(p1, p2); ORDERKEY_RSSIZE(p1, p2); ORDERKEY_MEM(p1, p2); return (0); } /* compare_swap - the comparison function for sorting by swap */ static int compare_swap(const void *arg1, const void *arg2) { struct kinfo_proc *p1 = *(struct kinfo_proc **)arg1; struct kinfo_proc *p2 = *(struct kinfo_proc **)arg2; ORDERKEY_SWAP(p1, p2); ORDERKEY_PCTCPU(p1, p2); ORDERKEY_CPTICKS(p1, p2); ORDERKEY_STATE(p1, p2); ORDERKEY_PRIO(p1, p2); ORDERKEY_RSSIZE(p1, p2); ORDERKEY_MEM(p1, p2); return (0); } /* assorted comparison functions for sorting by i/o */ int compare_iototal(const void *arg1, const void *arg2) { struct kinfo_proc *p1 = *(struct kinfo_proc **)arg1; struct kinfo_proc *p2 = *(struct kinfo_proc **)arg2; return (get_io_total(p2) - get_io_total(p1)); } static int compare_ioread(const void *arg1, const void *arg2) { struct kinfo_proc *p1 = *(struct kinfo_proc **)arg1; struct kinfo_proc *p2 = *(struct kinfo_proc **)arg2; long dummy, inp1, inp2; (void) get_io_stats(p1, &inp1, &dummy, &dummy, &dummy, &dummy); (void) get_io_stats(p2, &inp2, &dummy, &dummy, &dummy, &dummy); return (inp2 - inp1); } static int compare_iowrite(const void *arg1, const void *arg2) { struct kinfo_proc *p1 = *(struct kinfo_proc **)arg1; struct kinfo_proc *p2 = *(struct kinfo_proc **)arg2; long dummy, oup1, oup2; (void) get_io_stats(p1, &dummy, &oup1, &dummy, &dummy, &dummy); (void) get_io_stats(p2, &dummy, &oup2, &dummy, &dummy, &dummy); return (oup2 - oup1); } static int compare_iofault(const void *arg1, const void *arg2) { struct kinfo_proc *p1 = *(struct kinfo_proc **)arg1; struct kinfo_proc *p2 = *(struct kinfo_proc **)arg2; long dummy, flp1, flp2; (void) get_io_stats(p1, &dummy, &dummy, &flp1, &dummy, &dummy); (void) get_io_stats(p2, &dummy, &dummy, &flp2, &dummy, &dummy); return (flp2 - flp1); } static int compare_vcsw(const void *arg1, const void *arg2) { struct kinfo_proc *p1 = *(struct kinfo_proc **)arg1; struct kinfo_proc *p2 = *(struct kinfo_proc **)arg2; long dummy, flp1, flp2; (void) get_io_stats(p1, &dummy, &dummy, &dummy, &flp1, &dummy); (void) get_io_stats(p2, &dummy, &dummy, &dummy, &flp2, &dummy); return (flp2 - flp1); } int compare_ivcsw(const void *arg1, const void *arg2) { struct kinfo_proc *p1 = *(struct kinfo_proc **)arg1; struct kinfo_proc *p2 = *(struct kinfo_proc **)arg2; long dummy, flp1, flp2; (void) get_io_stats(p1, &dummy, &dummy, &dummy, &dummy, &flp1); (void) get_io_stats(p2, &dummy, &dummy, &dummy, &dummy, &flp2); return (flp2 - flp1); } /* * proc_owner(pid) - returns the uid that owns process "pid", or -1 if * the process does not exist. * It is EXTREMELY IMPORTANT that this function work correctly. * If top runs setuid root (as in SVR4), then this function * is the only thing that stands in the way of a serious * security problem. It validates requests for the "kill" * and "renice" commands. */ int proc_owner(int pid) { int cnt; struct kinfo_proc **prefp; struct kinfo_proc *pp; prefp = pref; cnt = pref_len; while (--cnt >= 0) { pp = *prefp++; if (pp->ki_pid == (pid_t)pid) return ((int)pp->ki_ruid); } return (-1); } static int swapmode(int *retavail, int *retfree) { int n; struct kvm_swap swapary[1]; static int pagesize = 0; static u_long swap_maxpages = 0; *retavail = 0; *retfree = 0; #define CONVERT(v) ((quad_t)(v) * pagesize / 1024) n = kvm_getswapinfo(kd, swapary, 1, 0); if (n < 0 || swapary[0].ksw_total == 0) return (0); if (pagesize == 0) pagesize = getpagesize(); if (swap_maxpages == 0) GETSYSCTL("vm.swap_maxpages", swap_maxpages); /* ksw_total contains the total size of swap all devices which may exceed the maximum swap size allocatable in the system */ if ( swapary[0].ksw_total > swap_maxpages ) swapary[0].ksw_total = swap_maxpages; *retavail = CONVERT(swapary[0].ksw_total); *retfree = CONVERT(swapary[0].ksw_total - swapary[0].ksw_used); n = (int)(swapary[0].ksw_used * 100.0 / swapary[0].ksw_total); return (n); } Index: head/usr.bin/top/top.1 =================================================================== --- head/usr.bin/top/top.1 (revision 334473) +++ head/usr.bin/top/top.1 (revision 334474) @@ -1,546 +1,547 @@ .\" NOTE: changes to the manual page for "top" should be made in the .\" file "top.X" and NOT in the file "top.1". .\" $FreeBSD$ .nr N -1 .nr D 2 .TH TOP 1 Local .UC 4 .SH NAME top \- display and update information about the top cpu processes .SH SYNOPSIS .B top [ .B \-abCHIijnPqStuvwz ] [ .BI \-d count ] [ .BI \-m io | cpu ] [ .BI \-o field ] [ .BI \-s time ] [ .BI \-J jail ] [ .BI \-U username ] [ .I number ] .SH DESCRIPTION .\" This defines appropriate quote strings for nroff and troff .ds lq \&" .ds rq \&" .if t .ds lq `` .if t .ds rq '' .\" Just in case these number registers aren't set yet... .if \nN==0 .nr N 10 .if \nD==0 .nr D 2 .I Top displays the top .if !\nN==-1 \nN processes on the system and periodically updates this information. .if \nN==-1 \ \{\ If standard output is an intelligent terminal (see below) then as many processes as will fit on the terminal screen are displayed by default. Otherwise, a good number of them are shown (around 20). .\} Raw cpu percentage is used to rank the processes. If .I number is given, then the top .I number processes will be displayed instead of the default. .PP .I Top makes a distinction between terminals that support advanced capabilities and those that do not. This distinction affects the choice of defaults for certain options. In the remainder of this document, an \*(lqintelligent\*(rq terminal is one that supports cursor addressing, clear screen, and clear to end of line. Conversely, a \*(lqdumb\*(rq terminal is one that does not support such features. If the output of .I top is redirected to a file, it acts as if it were being run on a dumb terminal. .SH OPTIONS .TP .B \-C Toggle CPU display mode. By default top displays the weighted CPU percentage in the WCPU column (this is the same value that .IR ps (1) displays as CPU). Each time .B \-C flag is passed it toggles between \*(lqraw cpu\*(rq mode and \*(lqweighted cpu\*(rq mode, showing the \*(lqCPU\*(rq or the \*(lqWCPU\*(rq column respectively. .TP .B \-S Show system processes in the display. Normally, system processes such as the pager and the swapper are not shown. This option makes them visible. .TP .B \-a Display command names derived from the argv[] vector, rather than real executable name. It's useful when you want to watch applications, that puts their status information there. If the real name differs from argv[0], it will be displayed in parenthesis. .TP .B \-b Use \*(lqbatch\*(rq mode. In this mode, all input from the terminal is ignored. Interrupt characters (such as ^C and ^\e) still have an effect. This is the default on a dumb terminal, or when the output is not a terminal. .TP .B \-H Display each thread for a multithreaded process individually. By default a single summary line is displayed for each process. .TP .B \-i Use \*(lqinteractive\*(rq mode. In this mode, any input is immediately read for processing. See the section on \*(lqInteractive Mode\*(rq for an explanation of which keys perform what functions. After the command is processed, the screen will immediately be updated, even if the command was not understood. This mode is the default when standard output is an intelligent terminal. .TP .B \-I Do not display idle processes. By default, top displays both active and idle processes. .TP .B \-j Display the .IR jail (8) ID. .TP .B \-t Do not display the .I top process. .TP .BI \-m display Display either 'cpu' or 'io' statistics. Default is 'cpu'. .TP .B \-n Use \*(lqnon-interactive\*(rq mode. This is identical to \*(lqbatch\*(rq mode. .TP .B \-P Display per-cpu CPU usage statistics. .TP .B \-q Renice .I top to -20 so that it will run faster. This can be used when the system is being very sluggish to improve the possibility of discovering the problem. This option can only be used by root. .TP .B \-u Do not take the time to map uid numbers to usernames. Normally, .I top will read as much of the file \*(lq/etc/passwd\*(rq as is necessary to map all the user id numbers it encounters into login names. This option disables all that, while possibly decreasing execution time. The uid numbers are displayed instead of the names. .TP .B \-v Write version number information to stderr then exit immediately. No other processing takes place when this option is used. To see current revision information while top is running, use the help command \*(lq?\*(rq. .TP .B \-w Display approximate swap usage for each process. .TP .B \-z Do not display the system idle process. .TP .BI \-d count Show only .I count displays, then exit. A display is considered to be one update of the screen. This option allows the user to select the number of displays he wants to see before .I top automatically exits. For intelligent terminals, no upper limit is set. The default is 1 for dumb terminals. Please, note that for .I count = 1 no information is available about the percentage of time spent by the CPU in every state. .TP .BI \-s time Set the delay between screen updates to .I time seconds. The default delay between updates is \nD seconds. .TP .BI \-o field Sort the process display area on the specified field. The field name is the name of the column as seen in the output, but in lower case: \*(lqcpu\*(lq, \*(rqsize\*(lq, \*(rqres\*(lq, \*(rqtime\*(lq, \*(rqpri\*(lq, \*(rqthreads\*(lq, \*(lqtotal\*(lq, \*(rqread\*(lq, \*(rqwrite\*(lq, \*(rqfault\*(lq, \*(rqvcsw\*(lq, \*(rqivcsw\*(lq, \*(lqjid\*(lq, \*(rqswap\*(lq or \*(rqpid\*(lq. .TP .BI \-J jail Show only those processes owned by .IR jail . This may be either the .B jid or .B name of the jail. Use .B 0 to limit to host processes. Using this option implies the .B \-j flag. .PP .BI \-U username Show only those processes owned by .IR username . This option currently only accepts usernames and will not understand uid numbers. .PP Both .I count and .I number fields can be specified as \*(lqinfinite\*(rq, indicating that they can stretch as far as possible. This is accomplished by using any proper prefix of the keywords \*(lqinfinity\*(rq, \*(lqmaximum\*(rq, or \*(lqall\*(rq. The default for .I count on an intelligent terminal is, in fact, .BI infinity . .PP The environment variable .B TOP is examined for options before the command line is scanned. This enables a user to set his or her own defaults. The number of processes to display can also be specified in the environment variable .BR TOP . The options .BR \-a , .BR \-C , .BR \-H , .BR \-I , .BR \-j , .BR \-P , .BR \-S , .BR \-t , .BR \-u , .BR \-w , and .B \-z are actually toggles. A second specification of any of these options will negate the first. Thus a user who has the environment variable .B TOP set to \*(lq\-I\*(rq may use the command \*(lqtop \-I\*(rq to see idle processes. .SH "INTERACTIVE MODE" When .I top is running in \*(lqinteractive mode\*(rq, it reads commands from the terminal and acts upon them accordingly. In this mode, the terminal is put in \*(lqCBREAK\*(rq, so that a character will be processed as soon as it is typed. Almost always, a key will be pressed when .I top is between displays; that is, while it is waiting for .I time seconds to elapse. If this is the case, the command will be processed and the display will be updated immediately thereafter (reflecting any changes that the command may have specified). This happens even if the command was incorrect. If a key is pressed while .I top is in the middle of updating the display, it will finish the update and then process the command. Some commands require additional information, and the user will be prompted accordingly. While typing this information in, the user's erase and kill keys (as set up by the command .IR stty ) are recognized, and a newline terminates the input. .PP These commands are currently recognized (^L refers to control-L): .TP .B ^L Redraw the screen. .IP "\fBh\fP\ or\ \fB?\fP" Display a summary of the commands (help screen). Version information is included in this display. .TP .B q Quit .IR top. .TP .B d Change the number of displays to show (prompt for new number). Remember that the next display counts as one, so typing .B d1 will make .I top show one final display and then immediately exit. .TP .B m Toggle the display between 'cpu' and 'io' modes. .TP .B n or # Change the number of processes to display (prompt for new number). .TP .B s Change the number of seconds to delay between displays (prompt for new number). .TP .B S Toggle the display of system processes. .TP .B a Toggle the display of process titles. .TP .B k Send a signal (\*(lqkill\*(rq by default) to a list of processes. This acts similarly to the command .IR kill (1)). .TP .B r Change the priority (the \*(lqnice\*(rq) of a list of processes. This acts similarly to the command .IR renice (8)). .TP .B u Display only processes owned by a specific set of usernames (prompt for username). If the username specified is simply \*(lq+\*(rq or \*(lq-\*(rq, then processes belonging to all users will be displayed. Usernames can be added to and removed from the set by prepending them with \*(lq+\*(rq and \*(lq-\*(rq, respectively. .TP .B o Change the order in which the display is sorted. This command is not available on all systems. The sort key names vary from system to system but usually include: \*(lqcpu\*(rq, \*(lqres\*(rq, \*(lqsize\*(rq, \*(lqtime\*(rq. The default is cpu. .TP .B e Display a list of system errors (if any) generated by the last .BR k ill or .BR r enice command. .TP .B H Toggle the display of threads. +Also toggles the display of PID or TID. .TP .B i (or .BR I ) Toggle the display of idle processes. .TP .B j Toggle the display of .IR jail (8) ID. .TP .B J Display only processes owned by a specific jail (prompt for jail). If the jail specified is simply \*(lq+\*(rq, then processes belonging to all jails and the host will be displayed. This will also enable the display of JID. .TP .B P Toggle the display of per-CPU statistics. .TP .B t Toggle the display of the .I top process. .TP .B w Toggle the display of swap usage. .TP .B z Toggle the display of the system idle process. .SH "THE DISPLAY" The actual display varies depending on the specific variant of Unix that the machine is running. This description may not exactly match what is seen by top running on this particular machine. Differences are listed at the end of this manual entry. .PP The top few lines of the display show general information about the state of the system, including the last process id assigned to a process (on most systems), the three load averages, the current time, the number of existing processes, the number of processes in each state (sleeping, running, starting, zombies, and stopped), and a percentage of time spent in each of the processor states (user, nice, system, and idle). It also includes information about physical and virtual memory allocation. .PP The remainder of the screen displays information about individual processes. This display is similar in spirit to .IR ps (1) but it is not exactly the same. PID is the process id, JID, when displayed, is the .IR jail (8) ID corresponding to the process, USERNAME is the name of the process's owner (if .B \-u is specified, a UID column will be substituted for USERNAME), PRI is the current priority of the process, NICE is the nice amount (in the range \-20 to 20), SIZE is the total size of the process (text, data, and stack), RES is the current amount of resident memory, SWAP is the approximate amount of swap, if enabled (SIZE, RES and SWAP are given in kilobytes), STATE is the current state (one of \*(lqSTART\*(rq, \*(lqRUN\*(rq (shown as \*(lqCPUn\*(rq on SMP systems), \*(lqSLEEP\*(rq, \*(lqSTOP\*(rq, \*(lqZOMB\*(rq, \*(lqWAIT\*(rq, \*(lqLOCK\*(rq or the event on which the process waits), C is the processor number on which the process is executing (visible only on SMP systems), TIME is the number of system and user cpu seconds that the process has used, WCPU, when displayed, is the weighted cpu percentage (this is the same value that .IR ps (1) displays as CPU), CPU is the raw percentage and is the field that is sorted to determine the order of the processes, and COMMAND is the name of the command that the process is currently running (if the process is swapped out, this column is marked \*(lq\*(rq). .SH NOTES If a process is in the \*(lqSLEEP\*(rq or \*(lqLOCK\*(rq state, the state column will report the name of the event or lock on which the process is waiting. Lock names are prefixed with an asterisk \*(lq*\*(rq while sleep events are not. .SH AUTHOR William LeFebvre, EECS Department, Northwestern University .SH ENVIRONMENT .DT TOP user-configurable defaults for options. .SH FILES .DT /dev/kmem kernel memory .br /dev/mem physical memory .br /etc/passwd used to map uid numbers to user names .br /boot/kernel/kernel system image .SH BUGS Don't shoot me, but the default for .B \-I has changed once again. So many people were confused by the fact that .I top wasn't showing them all the processes that I have decided to make the default behavior show idle processes, just like it did in version 2. But to appease folks who can't stand that behavior, I have added the ability to set \*(lqdefault\*(rq options in the environment variable .B TOP (see the OPTIONS section). Those who want the behavior that version 3.0 had need only set the environment variable .B TOP to \*(lq\-I\*(rq. .PP The command name for swapped processes should be tracked down, but this would make the program run slower. .PP As with .IR ps (1), things can change while .I top is collecting information for an update. The picture it gives is only a close approximation to reality. .SH SEE ALSO kill(1), ps(1), stty(1), mem(4), getrusage(2), renice(8) .SH "FreeBSD NOTES" .SH DESCRIPTION OF MEMORY Mem: 61M Active, 86M Inact, 368K Laundry, 22G Wired, 102G Free ARC: 15G Total, 9303M MFU, 6155M MRU, 1464K Anon, 98M Header, 35M Other 15G Compressed, 27G Uncompressed, 1.75:1 Ratio, 174M Overhead Swap: 4096M Total, 532M Free, 13% Inuse, 80K In, 104K Out .TP .B K: Kilobyte .TP .B M: Megabyte .TP .B G: Gigabyte .TP .B %: 1/100 .SS Physical Memory Stats .TP .B Active: number of bytes active .TP .B Inact: number of clean bytes inactive .TP .B Laundry: number of dirty bytes queued for laundering .TP .B Wired: number of bytes wired down, including BIO-level cached file data pages .TP .B Buf: number of bytes used for BIO-level disk caching .TP .B Free: number of bytes free .SS ZFS ARC Stats These stats are only displayed when the ARC is in use. .TP .B Total: number of wired bytes used for the ZFS ARC .TP .B MRU: number of ARC bytes holding most recently used data .TP .B MFU: number of ARC bytes holding most frequently used data .TP .B Anon: number of ARC bytes holding in flight data .TP .B Header: number of ARC bytes holding headers .TP .B Other: miscellaneous ARC bytes .TP .B Compressed: bytes of memory used by ARC caches .TP .B Uncompressed: bytes of data stored in ARC caches before compression .TP .B Ratio: compression ratio of data cached in the ARC .SS Swap Stats .TP .B Total: total available swap usage .TP .B Free: total free swap usage .TP .B Inuse: swap usage .TP .B In: bytes paged in from swap devices (last interval) .TP .B Out: bytes paged out to swap devices (last interval)