/* * pg_top - a top PostgreSQL users display for Unix * * SYNOPSIS: any Sun running SunOS version 4.x * * DESCRIPTION: * This is the machine-dependent module for SunOS 4.x. * This makes pg_top work on the following systems: * SunOS 4.0 * SunOS 4.0.1 * SunOS 4.0.2 (including 386i architecture) * SunOS 4.0.3 * SunOS 4.1 * SunOS 4.1.1 * SunOS 4.1.2 (including MP architectures) * SunOS 4.1.3 (including MP architectures) * SunOS 4.1.3_U1 (including MP architectures) * SunOS 4.1.4 (including MP architectures) * Solbourne OS/MP PRIOR to 4.1A * * LIBS: -lkvm * * CFLAGS: -DHAVE_GETOPT -DORDER * * AUTHOR: William LeFebvre * Solbourne support by David MacKenzie */ /* * #ifdef MULTIPROCESSOR means Sun MP. * #ifdef solbourne is for Solbourne. */ #include "config.h" #include #include /* make sure param.h gets loaded with KERNEL defined to get PZERO & NZERO */ #define KERNEL #include #undef KERNEL #include #include #include #include #include #include #include #include #include #include #include #include #ifdef solbourne #include #endif /* Older versions of SunOS don't have a typedef for pid_t. Hopefully this will catch all those cases without causing other problems. */ #ifndef __sys_stdtypes_h typedef int pid_t; #endif #include "pg_top.h" #include "machine.h" #include "utils.h" /* declarations for load_avg */ #include "loadavg.h" /* get_process_info passes back a handle. This is what it looks like: */ struct handle { struct 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)->p_time == 0 ? 0.0 : \ ((pct) / (1.0 - exp((pp)->p_time * logcpu)))) /* what we consider to be process size: */ #define PROCSIZE(pp) ((pp)->p_tsize + (pp)->p_dsize + (pp)->p_ssize) /* definitions for indices in the nlist array */ #define X_AVENRUN 0 #define X_CCPU 1 #define X_MPID 2 #define X_NPROC 3 #define X_PROC 4 #define X_TOTAL 5 #define X_CP_TIME 6 #define X_PAGES 7 #define X_EPAGES 8 static struct nlist nlst[] = { #ifdef i386 {"avenrun"}, /* 0 */ {"ccpu"}, /* 1 */ {"mpid"}, /* 2 */ {"nproc"}, /* 3 */ {"proc"}, /* 4 */ {"total"}, /* 5 */ {"cp_time"}, /* 6 */ {"pages"}, /* 7 */ {"epages"}, /* 8 */ #else {"_avenrun"}, /* 0 */ {"_ccpu"}, /* 1 */ {"_mpid"}, /* 2 */ {"_nproc"}, /* 3 */ {"_proc"}, /* 4 */ {"_total"}, /* 5 */ {"_cp_time"}, /* 6 */ {"_pages"}, /* 7 */ {"_epages"}, /* 8 */ #ifdef MULTIPROCESSOR {"_ncpu"}, #define X_NCPU 9 {"_xp_time"}, #define X_XP_TIME 10 #endif #endif {0} }; /* * These definitions control the format of the per-process area */ static char header[] = " PID X PRI NICE SIZE RES STATE TIME WCPU CPU COMMAND"; /* 0123456 -- field to fill in starts at header+6 */ #define UNAME_START 6 #define Proc_format \ "%5d %-8.8s %3d %4d %5s %5s %-5s %-6s %5.2f%% %5.2f%% %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 */ char *state_abbrev[] = { "", "sleep", "WAIT", "run\0\0\0", "start", "zomb", "stop" }; /* values that we stash away in _init and use in later routines */ static double logcpu; kvm_t *kd; /* these are retrieved from the kernel in _init */ static unsigned long proc; static int nproc; static load_avg ccpu; static unsigned long pages; static unsigned long epages; static int ncpu = 0; /* these are offsets obtained via nlist and used in the get_ functions */ static unsigned long mpid_offset; static unsigned long avenrun_offset; static unsigned long total_offset; static unsigned long cp_time_offset; #ifdef MULTIPROCESSOR static unsigned long xp_time_offset; #endif /* these are for calculating cpu state percentages */ static long cp_time[CPUSTATES]; static long cp_old[CPUSTATES]; static long cp_diff[CPUSTATES]; #ifdef MULTIPROCESSOR static long xp_time[NCPU][XPSTATES]; /* for now we only accumulate spin time, but extending this to pick up other stuff in xp_time is trivial. */ static long xp_old[NCPU]; #endif /* these are for detailing the process states */ int process_states[7]; char *procstatenames[] = { "", " sleeping, ", " ABANDONED, ", " running, ", " starting, ", " zombie, ", " stopped, ", NULL }; /* these are for detailing the cpu states */ int cpu_states[5]; char *cpustatenames[] = { "user", "nice", "system", "idle", #ifdef MULTIPROCESSOR "spin", #define XCP_SPIN 4 #endif NULL }; /* these are for detailing the memory statistics */ long memory_stats[4]; char *memorynames[] = { "K available, ", "K in use, ", "K free, ", "K locked", NULL }; /* these are names given to allowed sorting orders -- first is default */ char *ordernames[] = {"cpu", "size", "res", NULL}; /* forward definitions for comparison functions */ int compare_cpu(); int compare_size(); int compare_res(); int (*proc_compares[]) () = { compare_cpu, compare_size, compare_res, NULL }; /* these are for keeping track of the proc array */ static int bytes; static int pref_len; static struct proc *pbase; static struct proc **pref; /* these are for getting the memory statistics */ static struct page *physpage; static int bytesize; static int count; static int pageshift; /* log base 2 of the pagesize */ /* define pagetok in terms of pageshift */ #define pagetok(size) ((size) << pageshift) /* useful externals */ extern int errno; extern char *sys_errlist[]; long lseek(); long time(); machine_init(statics) struct statics *statics; { register int i; register int pagesize; /* initialize the kernel interface */ if ((kd = kvm_open(NULL, NULL, NULL, O_RDONLY, "pg_top")) == NULL) { perror("kvm_open"); return (-1); } /* get the list of symbols we want to access in the kernel */ if ((i = kvm_nlist(kd, nlst)) < 0) { fprintf(stderr, "pg_top: nlist failed\n"); return (-1); } #ifdef MULTIPROCESSOR /* were ncpu and xp_time not found in the nlist? */ if (i > 0 && nlst[X_NCPU].n_type == 0 && nlst[X_XP_TIME].n_type == 0) { /* we were compiled on an MP system but we are not running on one */ /* so we will pretend this didn't happen and set ncpu = 1 */ i -= 2; ncpu = 1; } #endif #ifdef solbourne { unsigned int status, type; /* Get the number of CPUs on this system. */ syscall(SYS_getcpustatus, &status, &ncpu, &type); } #endif /* make sure they were all found */ if (i > 0 && check_nlist(nlst) > 0) { return (-1); } /* get the symbol values out of kmem */ (void) getkval(nlst[X_PROC].n_value, (int *) (&proc), sizeof(proc), nlst[X_PROC].n_name); (void) getkval(nlst[X_NPROC].n_value, &nproc, sizeof(nproc), nlst[X_NPROC].n_name); (void) getkval(nlst[X_CCPU].n_value, (int *) (&ccpu), sizeof(ccpu), nlst[X_CCPU].n_name); (void) getkval(nlst[X_PAGES].n_value, (int *) (&pages), sizeof(pages), nlst[X_PAGES].n_name); (void) getkval(nlst[X_EPAGES].n_value, (int *) (&epages), sizeof(epages), nlst[X_EPAGES].n_name); #ifdef MULTIPROCESSOR if (ncpu == 0) { /* if ncpu > 0 then we are not really on an MP system */ (void) getkval(nlst[X_NCPU].n_value, (int *) (&ncpu), sizeof(ncpu), nlst[X_NCPU].n_name); } #endif /* stash away certain offsets for later use */ mpid_offset = nlst[X_MPID].n_value; avenrun_offset = nlst[X_AVENRUN].n_value; total_offset = nlst[X_TOTAL].n_value; cp_time_offset = nlst[X_CP_TIME].n_value; #ifdef MULTIPROCESSOR xp_time_offset = nlst[X_XP_TIME].n_value; #endif /* this is used in calculating WCPU -- calculate it ahead of time */ logcpu = log(loaddouble(ccpu)); /* allocate space for proc structure array and array of pointers */ bytes = nproc * sizeof(struct proc); pbase = (struct proc *) malloc(bytes); pref = (struct proc **) malloc(nproc * sizeof(struct proc *)); /* Just in case ... */ if (pbase == (struct proc *) NULL || pref == (struct proc **) NULL) { fprintf(stderr, "pg_top: can't allocate sufficient memory\n"); return (-1); } /* allocate a table to hold all the page structs */ bytesize = epages - pages; count = bytesize / sizeof(struct page); physpage = (struct page *) malloc(epages - pages); if (physpage == NULL) { fprintf(stderr, "pg_top: can't allocate sufficient memory\n"); return (-1); } /* get the page size with "getpagesize" 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; #if defined(MULTIPROCESSOR) || defined(solbourne) /* add a slash to the "run" state abbreviation */ if (ncpu > 1) { state_abbrev[SRUN][3] = '/'; } #endif /* fill in the statics information */ statics->procstate_names = procstatenames; statics->cpustate_names = cpustatenames; statics->memory_names = memorynames; statics->order_names = ordernames; /* all done! */ return (0); } char * format_header(uname_field) register char *uname_field; { register char *ptr; ptr = header + UNAME_START; while (*uname_field != '\0') { *ptr++ = *uname_field++; } return (header); } void get_system_info(si) struct system_info *si; { load_avg avenrun[3]; long total; #ifdef MULTIPROCESSOR long half_total; #endif /* get the cp_time array */ (void) getkval(cp_time_offset, (int *) cp_time, sizeof(cp_time), "_cp_time"); #ifdef MULTIPROCESSOR /* get the xp_time array as well */ if (ncpu > 1) { (void) getkval(xp_time_offset, (int *) xp_time, sizeof(xp_time), "_xp_time"); } #endif /* get load average array */ (void) getkval(avenrun_offset, (int *) avenrun, sizeof(avenrun), "_avenrun"); /* get mpid -- process id of last process */ (void) getkval(mpid_offset, &(si->last_pid), sizeof(si->last_pid), "_mpid"); /* get the array of physpage descriptors */ (void) getkval(pages, (int *) physpage, bytesize, "array _page"); /* convert load averages to doubles */ { register int i; register double *infoloadp; register load_avg *sysloadp; infoloadp = si->load_avg; sysloadp = avenrun; for (i = 0; i < 3; i++) { *infoloadp++ = loaddouble(*sysloadp++); } } /* convert cp_time counts to percentages */ total = percentages(CPUSTATES, cpu_states, cp_time, cp_old, cp_diff); #ifdef MULTIPROCESSOR /* calculate spin time from all processors */ if (ncpu > 1) { register int c; register int i; register long sum; register long change; /* collect differences for each processor and add them */ sum = 0; for (i = 0; i < ncpu; i++) { c = xp_time[i][XP_SPIN]; change = c - xp_old[i]; if (change < 0) { /* counter wrapped */ change = (long) ((unsigned long) c - (unsigned long) xp_old[i]); } sum += change; xp_old[i] = c; } /* * NOTE: I am assuming that the ticks found in xp_time are already * included in the ticks accumulated in cp_time. To get an accurate * reflection, therefore, we have to subtract the spin time from the * system time and recompute those two percentages. */ half_total = total / 2l; cp_diff[CP_SYS] -= sum; cpu_states[CP_SYS] = (int) ((cp_diff[CP_SYS] * 1000 + half_total) / total); cpu_states[XCP_SPIN] = (int) ((sum * 1000 + half_total) / total); } #endif /* sum memory statistics */ { register struct page *pp; register int cnt; register int inuse; register int free; register int locked; /* bop thru the array counting page types */ pp = physpage; inuse = free = locked = 0; for (cnt = count; --cnt >= 0; pp++) { if (pp->p_free) free++; else if (pp->p_lock || pp->p_keepcnt > 0) locked++; else inuse++; } /* convert memory stats to Kbytes */ memory_stats[0] = pagetok(inuse + free); memory_stats[1] = pagetok(inuse); memory_stats[2] = pagetok(free); memory_stats[3] = pagetok(locked); } /* set arrays and strings */ si->cpustates = cpu_states; si->memory = memory_stats; } static struct handle handle; caddr_t get_process_info(si, sel, compare_index) struct system_info *si; struct process_select *sel; int compare_index; { register int i; register int total_procs; register int active_procs; register struct proc **prefp; register struct proc *pp; /* these are copied out of sel for speed */ int show_idle; int show_system; int show_uid; int show_command; /* read all the proc structures in one fell swoop */ (void) getkval(proc, (int *) pbase, bytes, "proc array"); /* 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_system = sel->system; show_uid = sel->uid != -1; show_command = sel->command != NULL; /* count up process states and get pointers to interesting procs */ total_procs = 0; active_procs = 0; bzero((char *) process_states, sizeof(process_states)); prefp = pref; for (pp = pbase, i = 0; i < nproc; pp++, i++) { /* * Place pointers to each valid proc structure in pref[]. Process * slots that are actually in use have a non-zero status field. * Processes with SSYS set are system processes---these get ignored * unless show_sysprocs is set. */ if (pp->p_stat != 0 && (show_system || ((pp->p_flag & SSYS) == 0))) { total_procs++; process_states[pp->p_stat]++; if ((pp->p_stat != SZOMB) && (show_idle || (pp->p_pctcpu != 0) || (pp->p_stat == SRUN)) && (!show_uid || pp->p_uid == (uid_t) sel->uid)) { *prefp++ = pp; active_procs++; } } } /* if requested, sort the "interesting" processes */ qsort((char *) pref, active_procs, sizeof(struct proc *), proc_compares[compare_index]); /* remember active and total counts */ si->p_total = total_procs; si->p_active = pref_len = active_procs; /* pass back a handle */ handle.next_proc = pref; handle.remaining = active_procs; return ((caddr_t) & handle); } char fmt[MAX_COLS]; /* static area where result is built */ char * format_next_process(handle, get_userid) caddr_t handle; char *(*get_userid) (); { register struct proc *pp; register long cputime; register double pct; struct user u; struct handle *hp; /* find and remember the next proc structure */ hp = (struct handle *) handle; pp = *(hp->next_proc++); hp->remaining--; /* get the process's user struct and set cputime */ if (getu(pp, &u) == -1) { (void) strcpy(u.u_comm, ""); cputime = 0; } else { /* set u_comm for system processes */ if (u.u_comm[0] == '\0') { if (pp->p_pid == 0) { (void) strcpy(u.u_comm, "Swapper"); } else if (pp->p_pid == 2) { (void) strcpy(u.u_comm, "Pager"); } } cputime = u.u_ru.ru_utime.tv_sec + u.u_ru.ru_stime.tv_sec; } /* calculate the base for cpu percentages */ pct = pctdouble(pp->p_pctcpu); #ifdef MULTIPROCESSOR /* * If there is more than one cpu then add the processor number to the * "run/" string. Note that this will only show up if the process is in * the run state. Also note: when they start making Suns with more than * 9 processors this will break since the string will then be more than 5 * characters. */ if (ncpu > 1) { state_abbrev[SRUN][4] = (pp->p_cpuid & 0xf) + '0'; } #endif #ifdef solbourne if (ncpu > 1) { state_abbrev[SRUN][4] = (pp->p_lastcpu) + '0'; } #endif /* format this entry */ sprintf(fmt, Proc_format, pp->p_pid, (*get_userid) (pp->p_uid), pp->p_pri - PZERO, pp->p_nice - NZERO, format_k(pagetok(PROCSIZE(pp))), format_k(pagetok(pp->p_rssize)), state_abbrev[pp->p_stat], format_time(cputime), 100.0 * weighted_cpu(pct, pp), 100.0 * pct, printable(u.u_comm)); /* return the result */ return (fmt); } /* * getu(p, u) - get the user structure for the process whose proc structure * is pointed to by p. The user structure is put in the buffer pointed * to by u. Return 0 if successful, -1 on failure (such as the process * being swapped out). */ getu(p, u) register struct proc *p; struct user *u; { register struct user *lu; lu = kvm_getu(kd, p); if (lu == NULL) { return (-1); } else { *u = *lu; return (0); } } /* * check_nlist(nlst) - checks the nlist to see if any symbols were not * found. For every symbol that was not found, a one-line * message is printed to stderr. The routine returns the * number of symbols NOT found. */ int check_nlist(nlst) register struct nlist *nlst; { register int i; /* check to see if we got ALL the symbols we requested */ /* this will write one line to stderr for every symbol not found */ i = 0; while (nlst->n_name != NULL) { #ifdef i386 if (nlst->n_value == 0) #else if (nlst->n_type == 0) #endif { /* this one wasn't found */ fprintf(stderr, "kernel: no symbol named `%s'\n", nlst->n_name); i = 1; } nlst++; } return (i); } /* * getkval(offset, ptr, size, refstr) - get a value out of the kernel. * "offset" is the byte offset into the kernel for the desired value, * "ptr" points to a buffer into which the value is retrieved, * "size" is the size of the buffer (and the object to retrieve), * "refstr" is a reference string used when printing error meessages, * if "refstr" starts with a '!', then a failure on read will not * be fatal (this may seem like a silly way to do things, but I * really didn't want the overhead of another argument). * */ getkval(offset, ptr, size, refstr) unsigned long offset; int *ptr; int size; char *refstr; { if (kvm_read(kd, offset, ptr, size) != size) { if (*refstr == '!') { return (0); } else { fprintf(stderr, "pg_top: kvm_read for %s: %s\n", refstr, sys_errlist[errno]); quit(23); /* NOTREACHED */ } } return (1); } /* comparison routines for qsort */ /* * There are currently four possible comparison routines. main selects * one of these by indexing in to the array proc_compares. * * Possible keys are defined as macros below. Currently these keys are * defined: percent cpu, cpu ticks, process 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. */ /* First, the possible comparison keys. These are defined in such a way that they can be merely listed in the source code to define the actual desired ordering. */ #define ORDERKEY_PCTCPU if (lresult = p2->p_pctcpu - p1->p_pctcpu,\ (result = lresult > 0 ? 1 : lresult < 0 ? -1 : 0) == 0) #define ORDERKEY_CPTICKS if ((result = p2->p_cpticks - p1->p_cpticks) == 0) #define ORDERKEY_STATE if ((result = sorted_state[p2->p_stat] - \ sorted_state[p1->p_stat]) == 0) #define ORDERKEY_PRIO if ((result = p2->p_pri - p1->p_pri) == 0) #define ORDERKEY_RSSIZE if ((result = p2->p_rssize - p1->p_rssize) == 0) #define ORDERKEY_MEM if ((result = PROCSIZE(p2) - PROCSIZE(p1)) == 0) /* Now the array that maps process state to a weight */ static unsigned char sorted_state[] = { 0, /* not used */ 3, /* sleep */ 1, /* ABANDONED (WAIT) */ 6, /* run */ 5, /* start */ 2, /* zombie */ 4 /* stop */ }; /* compare_cpu - the comparison function for sorting by cpu percentage */ compare_cpu(pp1, pp2) struct proc **pp1; struct proc **pp2; { register struct proc *p1; register struct proc *p2; register int result; register pctcpu lresult; /* remove one level of indirection */ p1 = *pp1; p2 = *pp2; ORDERKEY_PCTCPU ORDERKEY_CPTICKS ORDERKEY_STATE ORDERKEY_PRIO ORDERKEY_RSSIZE ORDERKEY_MEM ; return (result); } /* compare_size - the comparison function for sorting by total memory usage */ compare_size(pp1, pp2) struct proc **pp1; struct proc **pp2; { register struct proc *p1; register struct proc *p2; register int result; register pctcpu lresult; /* remove one level of indirection */ p1 = *pp1; p2 = *pp2; ORDERKEY_MEM ORDERKEY_RSSIZE ORDERKEY_PCTCPU ORDERKEY_CPTICKS ORDERKEY_STATE ORDERKEY_PRIO ; return (result); } /* compare_res - the comparison function for sorting by resident set size */ compare_res(pp1, pp2) struct proc **pp1; struct proc **pp2; { register struct proc *p1; register struct proc *p2; register int result; register pctcpu lresult; /* remove one level of indirection */ p1 = *pp1; p2 = *pp2; ORDERKEY_RSSIZE ORDERKEY_MEM ORDERKEY_PCTCPU ORDERKEY_CPTICKS ORDERKEY_STATE ORDERKEY_PRIO ; return (result); } /* * proc_owner(pid) - returns the uid that owns process "pid", or -1 if * the process does not exist. * It is EXTREMLY IMPORTANT that this function work correctly. * If pg_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(pid) int pid; { register int cnt; register struct proc **prefp; register struct proc *pp; prefp = pref; cnt = pref_len; while (--cnt >= 0) { if ((pp = *prefp++)->p_pid == (pid_t) pid) { return ((int) pp->p_uid); } } return (-1); }