/*
* top - a top users display for Unix
*
* SYNOPSIS:  any hp9000 running hpux version 10.10 (ALPHA VERSION)
*
* DESCRIPTION:
* This is the machine-dependent module for HPUX 10 that uses pstat.
* This makes top work on (at least) the following systems:
*      hp9000s800
* This may make top work on the following, but we aren't sure:
*      hp9000s700
*      hp9000s300
*
* CFLAGS: -DHAVE_GETOPT
*
* LIBS:
*
* AUTHOR: Rich Holland <[email protected]>
* AUTHOR: adapted from Kevin Schmidt <[email protected]>
*/

#include <sys/types.h>
#include <sys/signal.h>
#include <sys/param.h>

#include <stdio.h>
#include <errno.h>
#include <nlist.h>
#include <math.h>
#include <sys/dir.h>
#include <sys/dk.h>
#include <sys/vm.h>
#include <sys/file.h>
#include <sys/time.h>
#include <sys/pstat.h>

# define P_RSSIZE(p) (p)->pst_rssize
# define P_TSIZE(p) (p)->pst_tsize
# define P_DSIZE(p) (p)->pst_dsize
# define P_SSIZE(p) (p)->pst_ssize

#include "top.h"
#include "machine.h"
#include "utils.h"

#define VMUNIX  "/stand/vmunix"
#define KMEM    "/dev/kmem"
#define MEM     "/dev/mem"
#ifdef DOSWAP
#define SWAP    "/dev/dmem"
#endif

/* get_process_info passes back a handle.  This is what it looks like: */

struct handle
{
   struct pst_status **next_proc;      /* points to next valid pst pointer */
   int remaining;              /* number of pointers remaining */
};

/* declarations for load_avg */
#include "loadavg.h"

/* define what weighted cpu is.  */
#define weighted_cpu(pct, pp) ((pp->pst_time) == 0 ? 0.0 : \
                        ((pct) / (1.0 - exp((pp->pst_time) * logcpu))))

/* what we consider to be process size: */
#define PROCSIZE(pp) (P_TSIZE(pp) + P_DSIZE(pp) + P_SSIZE(pp))

/* definitions for indices in the nlist array */
#define X_AVENRUN       0
#define X_CCPU          1
#define X_NPROC         2
#define X_PROC          3
#define X_TOTAL         4
#define X_CP_TIME       5
#define X_MPID          6

/*
* Steinar Haug from University of Trondheim, NORWAY pointed out that
* the HP 9000 system 800 doesn't have _hz defined in the kernel.  He
* provided a patch to work around this.  We've improved on this patch
* here and set the constant X_HZ only when _hz is available in the
* kernel.  Code in this module that uses X_HZ is surrounded with
* appropriate ifdefs.
*/

#ifndef hp9000s300
#define X_HZ            7
#endif


static struct nlist nlst[] = {
   { "_avenrun" },             /* 0 */
   { "_cexp" },                /* 1 */
   { "_nproc" },               /* 2 */
   { "_proc" },                /* 3 */
   { "_total" },               /* 4 */
   { "_cp_time" },             /* 5 */
   { "_mpid" },                /* 6 */
#ifdef X_HZ
   { "_hz" },                  /* 7 */
#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 */

char *state_abbrev[] =
{
   "", "sleep", "run", "stop", "zomb", "trans", "start"
};


static int kmem;

/* values that we stash away in _init and use in later routines */

static double logcpu;

/* these are retrieved from the kernel in _init */

static unsigned long proc;
static          int  nproc;
static          long hz;
static load_avg  ccpu;
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;

/* 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 */

int process_states[7];
char *procstatenames[] = {
   "", " sleeping, ", " running, ", " stopped, ", " zombie, ",
   " trans, ", " starting, ",
   NULL
};

/* these are for detailing the cpu states */

int cpu_states[9];
char *cpustatenames[] = {
   "usr", "nice", "sys", "idle", NULL, "", "", "intr", "ker",
   NULL
};

/* these are for detailing the memory statistics */

int memory_stats[8];
char *memorynames[] = {
   "Real: ", "K/", "K act/tot  ", "Virtual: ", "K/",
   "K act/tot  ", "Free: ", "K", NULL
};

/* these are for keeping track of the pst array */

static int bytes;
static int pref_len;
static struct pst_status *pbase;
static struct pst_status **pref;

/* these are for getting the memory statistics */

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 = 0;
   register int pagesize;

   if ((kmem = open(KMEM, O_RDONLY)) == -1) {
       perror(KMEM);
       return(-1);
   }
#ifdef hp9000s800
   /* 800 names don't have leading underscores */
   for (i = 0; nlst[i].n_name; nlst[i++].n_name++)
       continue;
#endif

   /* get the list of symbols we want to access in the kernel */
   (void) nlist(VMUNIX, nlst);
   if (nlst[0].n_type == 0)
   {
       fprintf(stderr, "top: nlist failed\n");
       return(-1);
   }

   /* make sure they were all found */
   if (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);
#ifdef X_HZ
   (void) getkval(nlst[X_HZ].n_value,     (int *)(&hz),        sizeof(hz),
           nlst[X_HZ].n_name);
#else
   hz = HZ;
#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;

   /* this is used in calculating WCPU -- calculate it ahead of time */
   logcpu = log(loaddouble(ccpu));

   /* allocate space for pst structure array and array of pointers */
   bytes = nproc * sizeof(struct pst_status);
   pbase = (struct pst_status *)malloc(bytes);
   pref  = (struct pst_status **)malloc(nproc * sizeof(struct pst_status *));

   /* Just in case ... */
   if (pbase == (struct pst_status *)NULL || pref == (struct pst_status **)NULL)
   {
       fprintf(stderr, "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;

   /* fill in the statics information */
   statics->procstate_names = procstatenames;
   statics->cpustate_names = cpustatenames;
   statics->memory_names = memorynames;

   /* 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);
}

get_system_info(si)

struct system_info *si;

{
   load_avg avenrun[3];
   long total;

   /* get the cp_time array */
   (void) getkval(cp_time_offset, (int *)cp_time, sizeof(cp_time),
                  "_cp_time");

   /* 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");

   /* 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);

   /* sum memory statistics */
   {
       struct vmtotal total;

       /* get total -- systemwide main memory usage structure */
       (void) getkval(total_offset, (int *)(&total), sizeof(total),
                      "_total");
       /* convert memory stats to Kbytes */
       memory_stats[0] = -1;
       memory_stats[1] = pagetok(total.t_arm);
       memory_stats[2] = pagetok(total.t_rm);
       memory_stats[3] = -1;
       memory_stats[4] = pagetok(total.t_avm);
       memory_stats[5] = pagetok(total.t_vm);
       memory_stats[6] = -1;
       memory_stats[7] = pagetok(total.t_free);
   }

   /* set arrays and strings */
   si->cpustates = cpu_states;
   si->memory = memory_stats;
}

static struct handle handle;

caddr_t get_process_info(si, sel, compare)

struct system_info *si;
struct process_select *sel;
int (*compare)();

{
   register int i;
   register int total_procs;
   register int active_procs;
   register struct pst_status **prefp;
   register struct pst_status *pp;

   /* these are copied out of sel for speed */
   int show_idle;
   int show_system;
   int show_uid;
   int show_command;

   /* read all the pst structures in one fell swoop */
   if ((total_procs = pstat_getproc(pbase, sizeof(struct pst_status),
                                    nproc, 0)) < 0)
   {
       perror("pstat_getproc");
       return(NULL);
   }

   /* 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 */
   active_procs = 0;
   memset((char *)process_states, 0, sizeof(process_states));
   prefp = pref;
   for (pp = pbase, i = 0; i < total_procs; pp++, i++)
   {
       /*
        *  Place pointers to each valid pst 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->pst_stat != 0 &&
           (show_system || ((pp->pst_flag & PS_SYS) == 0)))
       {
           process_states[pp->pst_stat]++;

           /*
            * idle processes can be selectively ignored:  a process is
            * considered idle when cpticks is zero AND it is not in the run
            * state.  Zombies are always ignored.  We also skip over
            * processes that have been excluded via a uid selection
            */
           if ((pp->pst_stat != PS_ZOMBIE) &&
               (show_idle || (pp->pst_cpticks != 0) || (pp->pst_stat == PS_RUN)) &&
               (!show_uid || pp->pst_uid == (uid_t)sel->uid))
           {
               *prefp++ = pp;
               active_procs++;
           }
       }
   }

   /* if requested, sort the "interesting" processes */
   if (compare != NULL)
   {
       qsort((char *)pref, active_procs, sizeof(struct pst_status *), compare);
   }

   /* 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 pst_status *pp;
   register long cputime;
   register double pct;
   int where;
   struct handle *hp;
   struct timeval time;
   struct timezone timezone;

   /* sanity check */
   if (handle == (caddr_t)NULL)
   {
       return("");
   }

   /* find and remember the next proc structure */
   hp = (struct handle *)handle;
   pp = *(hp->next_proc++);
   hp->remaining--;


   /* set ucomm for system processes */
   if (pp->pst_ucomm[0] == '\0')
   {
       /* is this ever going to happen? */
       if (pp->pst_pid == 0)
       {
           (void) strcpy(pp->pst_ucomm, "Swapper");
       }
       else if (pp->pst_pid == 2)
       {
           (void) strcpy(pp->pst_ucomm, "Pager");
       }
   }

   cputime = pp->pst_cptickstotal / hz;

   /* calculate the base for cpu percentages */
   pct = pctdouble(pp->pst_pctcpu);

   /* get time used for calculation in weighted_cpu */
   gettimeofday(&time, &timezone);

   /* format this entry */
   sprintf(fmt,
           Proc_format,
           pp->pst_pid,
           (*get_userid)(pp->pst_uid),
           pp->pst_pri,
           pp->pst_nice,
           format_k(pagetok(PROCSIZE(pp))),
           format_k(pagetok(P_RSSIZE(pp))),
           state_abbrev[pp->pst_stat],
           format_time(cputime),
           100.0 * weighted_cpu(pct, pp),
           100.0 * pct,
           printable(pp->pst_ucomm));

   /* return the result */
   return(fmt);
}

/*
* 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)
   {
       if (nlst->n_type == 0)
       {
           /* 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 (lseek(kmem, (long)offset, L_SET) == -1) {
       if (*refstr == '!')
           refstr++;
       (void) fprintf(stderr, "%s: lseek to %s: %s\n", KMEM,
                      refstr, strerror(errno));
       quit(23);
   }
   if (read(kmem, (char *) ptr, size) == -1) {
       if (*refstr == '!')
           return(0);
       else {
           (void) fprintf(stderr, "%s: reading %s: %s\n", KMEM,
                          refstr, strerror(errno));
           quit(23);
       }
   }
   return(1);
}

/* comparison routine for qsort */

/*
*  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 unsigned char sorted_state[] =
{
   0,  /* not used             */
   3,  /* sleep                */
   6,  /* run                  */
   4,  /* stop                 */
   2,  /* zombie               */
   5,  /* start                */
   1,  /* other                */
};

proc_compare(pp1, pp2)

struct pst_status **pp1;
struct pst_status **pp2;

{
   register struct pst_status *p1;
   register struct pst_status *p2;
   register int result;
   register pctcpu lresult;

   /* remove one level of indirection */
   p1 = *pp1;
   p2 = *pp2;

   /* compare percent cpu (pctcpu) */
   if ((lresult = p2->pst_pctcpu - p1->pst_pctcpu) == 0)
   {
       /* use cpticks to break the tie */
       if ((result = p2->pst_cpticks - p1->pst_cpticks) == 0)
       {
           /* use process state to break the tie */
           if ((result = sorted_state[p2->pst_stat] -
                         sorted_state[p1->pst_stat])  == 0)
           {
               /* use priority to break the tie */
               if ((result = p2->pst_pri - p1->pst_pri) == 0)
               {
                   /* use resident set size (rssize) to break the tie */
                   if ((result = P_RSSIZE(p2) - P_RSSIZE(p1)) == 0)
                   {
                       /* use total memory to break the tie */
                       result = PROCSIZE(p2) - PROCSIZE(p1);
                   }
               }
           }
       }
   }
   else
   {
       result = lresult < 0 ? -1 : 1;
   }

   return(result);
}


void (*signal(sig, func))()
   int sig;
   void (*func)();
{
   struct sigvec osv, sv;

   /*
    * XXX: we should block the signal we are playing with,
    *      in case we get interrupted in here.
    */
   if (sigvector(sig, NULL, &osv) == -1)
       return BADSIG;
   sv = osv;
   sv.sv_handler = func;
#ifdef SV_BSDSIG
   sv.sv_flags |= SV_BSDSIG;
#endif
   if (sigvector(sig, &sv, NULL) == -1)
       return BADSIG;
   return osv.sv_handler;
}

int getpagesize() { return 1 << PGSHIFT; }

int setpriority(a, b, c) { errno = ENOSYS; return -1; }

/*
* 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 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 pst_status **prefp;
   register struct pst_status *pp;

   prefp = pref;
   cnt = pref_len;
   while (--cnt >= 0)
   {
       if ((pp = *prefp++)->pst_pid == (pid_t)pid)
       {
           return((int)pp->pst_uid);
       }
   }
   return(-1);
}