/* * OpenPBS (Portable Batch System) v2.3 Software License * * Copyright (c) 1999-2000 Veridian Information Solutions, Inc. * All rights reserved. * * --------------------------------------------------------------------------- * For a license to use or redistribute the OpenPBS software under conditions * other than those described below, or to purchase support for this software, * please contact Veridian Systems, PBS Products Department ("Licensor") at: * * www.OpenPBS.org +1 650 967-4675 sales@OpenPBS.org * 877 902-4PBS (US toll-free) * --------------------------------------------------------------------------- * * This license covers use of the OpenPBS v2.3 software (the "Software") at * your site or location, and, for certain users, redistribution of the * Software to other sites and locations. Use and redistribution of * OpenPBS v2.3 in source and binary forms, with or without modification, * are permitted provided that all of the following conditions are met. * After December 31, 2001, only conditions 3-6 must be met: * * 3. Any Redistribution of source code must retain the above copyright notice * and the acknowledgment contained in paragraph 6, this list of conditions * and the disclaimer contained in paragraph 7. * * 4. Any Redistribution in binary form must reproduce the above copyright * notice and the acknowledgment contained in paragraph 6, this list of * conditions and the disclaimer contained in paragraph 7 in the * documentation and/or other materials provided with the distribution. * * 5. Redistributions in any form must be accompanied by information on how to * obtain complete source code for the OpenPBS software and any * modifications and/or additions to the OpenPBS software. The source code * must either be included in the distribution or be available for no more * than the cost of distribution plus a nominal fee, and all modifications * and additions to the Software must be freely redistributable by any party * (including Licensor) without restriction. * * 6. All advertising materials mentioning features or use of the Software must * display the following acknowledgment: * * "This product includes software developed by NASA Ames Research Center, * Lawrence Livermore National Laboratory, and Veridian Information * Solutions, Inc. * Visit www.OpenPBS.org for OpenPBS software support, * products, and information." * * 7. DISCLAIMER OF WARRANTY * * THIS SOFTWARE IS PROVIDED "AS IS" WITHOUT WARRANTY OF ANY KIND. ANY EXPRESS * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, AND NON-INFRINGEMENT * ARE EXPRESSLY DISCLAIMED. * * IN NO EVENT SHALL VERIDIAN CORPORATION, ITS AFFILIATED COMPANIES, OR THE * U.S. GOVERNMENT OR ANY OF ITS AGENCIES BE LIABLE FOR ANY DIRECT OR INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, * OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, * EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * * This license will be governed by the laws of the Commonwealth of Virginia, * without reference to its choice of law rules. */ #include /* the master config generated by configure */ #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 #include "quota.h" /* exteed of and */ #include "pbs_error.h" #include "portability.h" #include "list_link.h" #include "server_limits.h" #include "attribute.h" #include "resource.h" #include "pbs_job.h" #include "log.h" #include "mom_mach.h" #include "mom_func.h" #include "resmon.h" #include "utils.h" #include "../rm_dep.h" #include #include #include #include #include /* ** System dependent code to gather information for the resource ** monitor for a Linux i386 machine. ** ** Resources known by this code: ** cput cpu time for a pid or session ** mem memory size for a pid or session in KB ** resi resident memory size for a pid or session in KB ** sessions list of sessions in the system ** pids list of pids in a session ** nsessions number of sessions in the system ** nusers number of users in the system ** totmem total memory size in KB ** availmem available memory size in KB ** ncpus number of cpus ** physmem physical memory size in KB ** size size of a file or filesystem ** idletime seconds of idle time ** walltime wall clock time for a pid ** loadave current load average ** quota quota information (sizes in kb) ** netload number of bytes transferred for all interfaces */ #ifndef TRUE #define FALSE 0 #define TRUE 1 #endif /* TRUE */ /***************************************************/ /*** setrlimit don't support on CYGWIN till now ***/ /***************************************************/ #undef HAVE_GYGWIN_SETRLIMIT static char procfs[] = "/proc"; static DIR *pdir = NULL; static int pagesize; extern char *ret_string; extern char extra_parm[]; extern char no_parm[]; extern time_t time_now; extern int LOGLEVEL; extern char PBSNodeMsgBuf[1024]; #define TBL_INC 200 /* initial proc table */ static proc_stat_t *proc_array = NULL; static int nproc = 0; static int max_proc = 0; /* ** external functions and data */ extern struct config *search (struct config *,char *); extern struct rm_attribute *momgetattr (char *); extern int rm_errno; extern double cputfactor; extern double wallfactor; extern long system_ncpus; extern int ignwalltime; extern int igncput; extern int ignvmem; extern int ignmem; /* ** local functions and data */ static char *resi (struct rm_attribute *); static char *totmem (struct rm_attribute *); static char *availmem (struct rm_attribute *); static char *physmem (struct rm_attribute *); static char *ncpus (struct rm_attribute *); static char *walltime (struct rm_attribute *); static char *quota (struct rm_attribute *); static char *netload (struct rm_attribute *); #ifndef mbool_t #define mbool_t char #endif /* mbool_t */ mbool_t ProcIsChild(char *,char *,char *); extern char *loadave (struct rm_attribute *); extern char *nullproc (struct rm_attribute *); time_t wait_time = 10; typedef struct proc_mem { unsigned long long mem_total; unsigned long long mem_used; unsigned long long mem_free; unsigned long long swap_total; unsigned long long swap_used; unsigned long long swap_free; } proc_mem_t; /* ** local resource array */ struct config dependent_config[] = { { "resi", {resi} }, { "totmem", {totmem} }, { "availmem", {availmem} }, { "physmem", {physmem} }, { "ncpus", {ncpus} }, { "loadave", {loadave} }, { "walltime", {walltime} }, { "quota", {quota} }, { "netload", {netload} }, { "size", {size} }, { NULL, {nullproc} } }; unsigned linux_time = 0; /* * support routine for getting system time -- sets linux_time */ void proc_get_btime(void) { FILE *fp; char label[256]; if ((fp = fopen("/proc/stat", "r")) == NULL) { return; } while (!feof(fp)) { if (fscanf(fp, "%s", label) != 1) { fclose(fp); return; } if (strcmp(label, "btime")) { if (fscanf(fp, "%*[^\n]%*c") != 0) { fclose(fp); return; } } else { if (fscanf(fp, "%u", &linux_time) != 1) {} fclose(fp); return; } } /* END while (!feof(fp)) */ fclose(fp); return; } /* END proc_get_btime() */ /* NOTE: see 'man 5 proc' for /proc/pid/stat format and description */ /* NOTE: leading '*' indicates that field should be ignored */ /* FORMAT: [] [] [] [] [] [] [] [] [<0>] [] [] [] ... */ static char stat_str[] = " %c %d %d %d %*d %*d %u %*u \ %*u %*u %*u %lu %lu %lu %lu %*ld %*ld %*u %*ld %lu %llu %lld %*lu %*lu \ %*lu %*lu %*lu %*lu %*lu %*lu %*lu %*lu %*lu %*lu %*lu"; /* * Convert jiffies to seconds. * * Hertz is sysconf(_SC_CLK_TCK) in get_proc_stat() */ #define JTOS(x) (x) / Hertz; /* * Linux /proc status routine. * * Returns a pointer to a static proc_stat_t structure given * a process number, or NULL if there is an error. Takes the * place of the ioctl call PIOCSTATUS in the irix imp of mom_mach.c * */ proc_stat_t *get_proc_stat( int pid) /* I */ { static proc_stat_t ps; static char path[1024]; static char readbuf[4096]; static char *lastbracket; FILE *fd; unsigned long jstarttime; /* number of jiffies since OS start time when process started */ struct stat sb; static int Hertz = 0; int Hertz_errored = 0; if (Hertz <= 0) { Hertz = sysconf(_SC_CLK_TCK); /* returns 0 on error */ if (Hertz <= 0) { /* FAILURE */ if (!Hertz_errored) log_err(errno, "get_proc_stat", "sysconf(_SC_CLK_TCK) failed, unable to monitor processes"); Hertz_errored = 1; return(NULL); } } Hertz_errored = 0; sprintf(path, "/proc/%d/stat", pid); if ((fd = fopen(path, "r")) == NULL) { /* FAILURE */ return(NULL); } /* use 'man 5 proc' for /proc/pid/stat format */ if (!fgets(readbuf, sizeof(readbuf), fd)) { fclose(fd); return(NULL); } lastbracket = strrchr(readbuf, ')'); if (lastbracket == NULL) { fclose(fd); return(NULL); } *lastbracket = '\0'; /* We basically split the string here, overwriting the ')'. */ lastbracket++; if (sscanf(readbuf,"%d (%[^\n]",&ps.pid,path) != 2) { /* FAILURE */ fclose(fd); return(NULL); } /* see stat_str[] value for mapping 'stat' format */ if (sscanf(lastbracket,stat_str, &ps.state, /* state (one of RSDZTW) */ &ps.ppid, /* ppid */ &ps.pgrp, /* pgrp */ &ps.session, /* session id */ &ps.flags, /* flags - kernel flags of the process, see the PF_* in */ &ps.utime, /* utime - jiffies that this process has been scheduled in user mode */ &ps.stime, /* stime - jiffies that this process has been scheduled in kernel mode */ &ps.cutime, /* cutime - jiffies that this process’s waited-for children have been scheduled in user mode */ &ps.cstime, /* cstime - jiffies that this process’s waited-for children have been scheduled in kernel mode */ &jstarttime, /* starttime */ &ps.vsize, /* vsize */ &ps.rss) != 12) /* rss */ { /* FAILURE */ fclose(fd); return(NULL); } if (fstat(fileno(fd), &sb) == -1) { /* FAILURE */ fclose(fd); return(NULL); } ps.uid = sb.st_uid; ps.start_time = linux_time + JTOS(jstarttime); ps.name = path; ps.utime = JTOS(ps.utime); ps.stime = JTOS(ps.stime); ps.cutime = JTOS(ps.cutime); ps.cstime = JTOS(ps.cstime); /* SUCCESS */ fclose(fd); return(&ps); } /* END get_proc_stat() */ proc_mem_t *get_proc_mem(void) { static proc_mem_t mm; FILE *fp; char str[32]; if ((fp = fopen("/proc/meminfo","r")) == NULL) { return(NULL); } if (fscanf(fp,"%30s",str) != 1) { return(NULL); } if (!strncmp(str,"total:",sizeof(str))) { /* old format */ if (fscanf(fp,"%*[^\n]%*c") != 0) /* remove text header */ { return(NULL); } /* umu vmem patch */ if (fscanf(fp, "%*s %llu %llu %llu %*u", &mm.mem_total, &mm.mem_used, &mm.mem_free) != 3) { return(NULL); } /* if (fscanf(fp,"%*s %lu %lu %lu %*[^\n]%*c", &mm.mem_total, &mm.mem_used, &mm.mem_free) != 3) { return(NULL); } */ if (fscanf(fp, "%*s %llu %llu %llu %*[^\n]%*c", &mm.swap_total, &mm.swap_used, &mm.swap_free) != 3) { return(NULL); } } fclose(fp); return(&mm); } /* END get_proc_mem() */ void dep_initialize(void) { char *id = "dep_initialize"; pagesize = getpagesize(); if ((pdir = opendir(procfs)) == NULL) { log_err(errno, id, "opendir"); return; } proc_get_btime(); return; } /* END dep_initialize() */ void dep_cleanup(void) { char *id = "dep_cleanup"; log_record(PBSEVENT_SYSTEM, 0, id, "dependent cleanup"); if (pdir) { closedir(pdir); pdir = NULL; } return; } /* * This routine is called on each cycle of the main loop. */ void dep_main_loop_cycle(void) { /* No periodic functions. */ } /* * Internal size decoding routine. * * Accepts a resource pointer and a pointer to the unsigned long integer * to receive the decoded value. It returns a PBS error code, and the * decoded value in the unsigned long integer. * * sizeof(word) = sizeof(int) */ static int getsize( resource *pres, /* I */ unsigned long *ret) /* O */ { unsigned long value; if (pres->rs_value.at_type != ATR_TYPE_SIZE) { return(PBSE_ATTRTYPE); } value = pres->rs_value.at_val.at_size.atsv_num; if (pres->rs_value.at_val.at_size.atsv_units == ATR_SV_WORDSZ) { if (value > ULONG_MAX / sizeof(int)) { return(PBSE_BADATVAL); } value *= sizeof(int); } if (value > (ULONG_MAX >> pres->rs_value.at_val.at_size.atsv_shift)) { return(PBSE_BADATVAL); } *ret = (value << pres->rs_value.at_val.at_size.atsv_shift); return(PBSE_NONE); } /* END getsize() */ /* * Internal time decoding routine. * * Accepts a resource pointer and a pointer to the unsigned long integer * to receive the decoded value. It returns a PBS error code, and the * decoded value of time in seconds in the unsigned long integer. */ static int gettime( resource *pres, unsigned long *ret) { if (pres->rs_value.at_type != ATR_TYPE_LONG) { return(PBSE_ATTRTYPE); } if (pres->rs_value.at_val.at_long < 0) { return(PBSE_BADATVAL); } *ret = pres->rs_value.at_val.at_long; return(PBSE_NONE); } static int injob( job *pjob, pid_t sid) { task *ptask; for (ptask = (task *)GET_NEXT(pjob->ji_tasks); ptask != NULL; ptask = (task *)GET_NEXT(ptask->ti_jobtask)) { if (ptask->ti_qs.ti_sid <= 1) continue; if (ptask->ti_qs.ti_sid == sid) { return(TRUE); } } return(FALSE); } /* END injob() */ /* * Internal session CPU time decoding routine. * * Accepts a job pointer. Returns the sum of all cpu time * consumed for all tasks executed by the job, in seconds, * adjusted by cputfactor. */ static unsigned long cput_sum( job *pjob) /* I */ { char *id = "cput_sum"; ulong cputime; int nps = 0; int i; proc_stat_t *ps; cputime = 0; if (LOGLEVEL >= 6) { sprintf(log_buffer, "proc_array loop start - jobid = %s", pjob->ji_qs.ji_jobid); log_record(PBSEVENT_DEBUG, 0, id, log_buffer); } for (i = 0;i < nproc;i++) { ps = &proc_array[i]; if ((LOGLEVEL >= 6) && (ps == NULL)) { sprintf(log_buffer, "proc_array loop end - nproc=%d, i=%d, ps is null", nproc, i); log_record(PBSEVENT_DEBUG, 0, id, log_buffer); } if (!injob(pjob, ps->session)) continue; nps++; cputime += (ps->utime + ps->stime + ps->cutime + ps->cstime); if (LOGLEVEL >= 6) { sprintf(log_buffer, "%s: session=%d pid=%d cputime=%lu (cputfactor=%f)", id, ps->session, ps->pid, cputime, cputfactor); log_record(PBSEVENT_SYSTEM, 0, id, log_buffer); } } /* END for (i) */ if (nps == 0) pjob->ji_flags |= MOM_NO_PROC; else pjob->ji_flags &= ~MOM_NO_PROC; return((unsigned long)((double)cputime * cputfactor)); } /* END cput_sum() */ /* * Return TRUE if any process in the job is over limit for cputime usage. */ static int overcpu_proc( job *pjob, unsigned long limit) /* I */ { char *id = "overcpu_proc"; struct dirent *dent; ulong cputime; proc_stat_t *ps; rewinddir(pdir); while ((dent = readdir(pdir)) != NULL) { if (!isdigit(dent->d_name[0])) continue; if ((ps = get_proc_stat(atoi(dent->d_name))) == NULL) { if (errno != ENOENT) { sprintf(log_buffer, "%s: get_proc_stat", dent->d_name); log_err(errno, id, log_buffer); } continue; } if (!injob(pjob, ps->session)) continue; /* change from ps->cutime to ps->utime, and ps->cstime to ps->stime */ cputime = (ulong)((double)(ps->utime + ps->stime) * cputfactor); if (cputime > limit) { return(TRUE); } } return(FALSE); } /* END overcpu_proc() */ /* * Internal session memory usage function. * * Returns the total number of bytes of address * space consumed by all current processes within the job. */ /* NOTE: routine should be modified to return llu */ static unsigned long long mem_sum( job *pjob) { char *id = "mem_sum"; int i; unsigned long long segadd; proc_stat_t *ps; segadd = 0; if (LOGLEVEL >= 6) { sprintf(log_buffer, "proc_array loop start - jobid = %s", pjob->ji_qs.ji_jobid); log_record(PBSEVENT_DEBUG, 0, id, log_buffer); } for (i = 0;i < nproc;i++) { ps = &proc_array[i]; if (!injob(pjob, ps->session)) continue; segadd += ps->vsize; } /* END for (i) */ return(segadd); } /* END mem_sum() */ /* * Internal session workingset size function. */ static unsigned long resi_sum( job *pjob) { char *id = "resi_sum"; ulong resisize; int i; proc_stat_t *ps; resisize = 0; if (LOGLEVEL >= 6) { sprintf(log_buffer, "proc_array loop start - jobid = %s", pjob->ji_qs.ji_jobid); log_record(PBSEVENT_DEBUG, 0, id, log_buffer); } for (i = 0;i < nproc;i++) { ps = &proc_array[i]; if (!injob(pjob, ps->session)) continue; resisize += ps->rss * pagesize; } /* END for (i) */ return(resisize); } /* END resi_sum() */ /* * Return TRUE if any process in the job is over limit for memory usage. */ static int overmem_proc( job *pjob, /* I */ unsigned long long limit) /* I */ { char *id = "overmem_proc"; int i; proc_stat_t *ps; if (LOGLEVEL >= 6) { sprintf(log_buffer, "proc_array loop start - jobid = %s", pjob->ji_qs.ji_jobid); log_record(PBSEVENT_DEBUG, 0, id, log_buffer); } for (i = 0;i < nproc;i++) { ps = &proc_array[i]; if (!injob(pjob, ps->session)) continue; if (ps->vsize > limit) { return(TRUE); } } /* END for (i) */ return(FALSE); } /* END overmem_proc() */ extern char *msg_momsetlim; /* * Internal error routine */ int error( char *string, int value) { char *message; assert(string != NULL); assert(*string != '\0'); message = pbse_to_txt(value); assert(message != NULL); assert(*message != '\0'); fprintf(stderr, msg_momsetlim, string, message); fflush(stderr); return(value); } /* END error() */ /* * Establish system-enforced limits for the job. * * Run through the resource list, checking the values for all items * we recognize. * * If set_mode is SET_LIMIT_SET, then also set hard limits for the * system enforced limits (not-polled). * If anything goes wrong with the process, return a PBS error code * and print a message on standard error. A zero-length resource list * is not an error. * * If set_mode is SET_LIMIT_SET the entry conditions are: * 1. MOM has already forked, and we are called from the child. * 2. The child is still running as root. * 3. Standard error is open to the user's file. * * If set_mode is SET_LIMIT_ALTER, we are being called to modify * existing limits. Cannot alter those set by setrlimit (kernel) * because we are the wrong process. */ int mom_set_limits( job *pjob, /* I */ int set_mode) /* SET_LIMIT_SET or SET_LIMIT_ALTER */ { char *id = "mom_set_limits"; char *pname = NULL; int retval; unsigned long value; /* place in which to build resource value */ resource *pres; struct rlimit reslim; unsigned long vmem_limit = 0; unsigned long mem_limit = 0; /* NOTE: log_buffer is exported */ if (LOGLEVEL >= 2) { sprintf(log_buffer, "%s(%s,%s) entered", id, (pjob != NULL) ? pjob->ji_qs.ji_jobid : "NULL", (set_mode == SET_LIMIT_SET) ? "set" : "alter"); log_record(PBSEVENT_SYSTEM, 0, id, log_buffer); log_buffer[0] = '\0'; } assert(pjob != NULL); assert(pjob->ji_wattr[(int)JOB_ATR_resource].at_type == ATR_TYPE_RESC); pres = (resource *)GET_NEXT(pjob->ji_wattr[(int)JOB_ATR_resource].at_val.at_list); /* * cycle through all the resource specifications, * setting limits appropriately. */ memset(&reslim, 0, sizeof(reslim)); while (pres != NULL) { if (pres->rs_defin != NULL) pname = pres->rs_defin->rs_name; else pname = NULL; if (LOGLEVEL >= 2) { sprintf(log_buffer, "setting limit for attribute '%s'", (pname != NULL) ? pname : "NULL"); log_record(PBSEVENT_SYSTEM, 0, id, log_buffer); log_buffer[0] = '\0'; } assert(pres->rs_defin != NULL); assert(pname != NULL); assert(pname[0] != '\0'); if (!strcmp(pname, "cput")) { if (igncput == FALSE) { /* cpu time - check, if less than pcput use it */ retval = gettime(pres, &value); if (retval != PBSE_NONE) { sprintf(log_buffer, "cput gettime failed in %s",id); return(error(pname, retval)); } } } else if (!strcmp(pname, "pcput")) { if (igncput == FALSE) { if (set_mode == SET_LIMIT_SET) { /* process cpu time - set */ retval = gettime(pres, &value); if (retval != PBSE_NONE) { sprintf(log_buffer, "pcput gettime failed in %s",id); return(error(pname, retval)); } reslim.rlim_cur = reslim.rlim_max = (unsigned long)((double)value / cputfactor); if (LOGLEVEL >= 2) { sprintf(log_buffer, "setting cpu time limit to %ld for job %s", (long int)reslim.rlim_cur, pjob->ji_qs.ji_jobid); log_record(PBSEVENT_SYSTEM, 0, id, log_buffer); log_buffer[0] = '\0'; } #ifdef HAVE_GYGWIN_SETRLIMIT if (setrlimit(RLIMIT_CPU, &reslim) < 0) { sprintf(log_buffer, "setrlimit for RLIMIT_CPU failed in %s, errno=%d (%s)", id, errno, strerror(errno)); return(error("RLIMIT_CPU", PBSE_SYSTEM)); } #endif /* HAVE_GYGWIN_SETRLIMIT */ } /* END if (set_mode == SET_LIMIT_SET) */ } } else if (!strcmp(pname, "file")) { /* set */ if (set_mode == SET_LIMIT_SET) { retval = getsize(pres, &value); if (retval != PBSE_NONE) { sprintf(log_buffer, "getsize() failed for file in %s", id); return(error(pname, retval)); } if (value > ULONG_MAX) { if (LOGLEVEL >= 0) { sprintf(log_buffer, "cannot set file limit to %ld for job %s (value too large)", (long int)reslim.rlim_cur, pjob->ji_qs.ji_jobid); log_err(-1, id, log_buffer); log_buffer[0] = '\0'; } return(error(pname, PBSE_BADATVAL)); } reslim.rlim_cur = reslim.rlim_max = value; #ifdef HAVE_GYGWIN_SETRLIMIT if (setrlimit(RLIMIT_FSIZE, &reslim) < 0) { sprintf(log_buffer, "cannot set file limit to %ld for job %s (setrlimit failed - check default user limits)", (long int)reslim.rlim_max, pjob->ji_qs.ji_jobid); log_err(errno, id, log_buffer); log_buffer[0] = '\0'; return(error(pname, PBSE_SYSTEM)); } #endif /* HAVE_GYGWIN_SETRLIMIT */ } } else if (!strcmp(pname, "vmem")) { if (ignvmem == FALSE) { /* check */ retval = getsize(pres, &value); if (retval != PBSE_NONE) { sprintf(log_buffer, "getsize() failed for vmem in %s",id); return(error(pname, retval)); } if ((vmem_limit == 0) || (value < vmem_limit)) vmem_limit = value; } } else if (!strcmp(pname, "pvmem")) { if (ignvmem == FALSE) { /* set */ if (set_mode == SET_LIMIT_SET) { retval = getsize(pres, &value); if (retval != PBSE_NONE) { sprintf(log_buffer, "getsize() failed for pvmem in %s", id); return(error(pname, retval)); } if (value > ULONG_MAX) { log_buffer[0] = '\0'; sprintf(log_buffer, "invalid value returned by getsize() for pvmem in %s", id); return(error(pname, PBSE_BADATVAL)); } if ((vmem_limit == 0) || (value < vmem_limit)) vmem_limit = value; } } } else if ((!strcmp(pname,"mem") && (pjob->ji_numnodes != 1)) || !strcmp(pname,"mppmem")) { /* ignore. If we ever get rid of support for the UNICOS OS then we can remove the ATR_DFLAG_MOM | ATR_DFLAG_ALTRUN flags from mppmem */ } else if ((!strcmp(pname, "mem") && (pjob->ji_numnodes == 1)) || !strcmp(pname, "pmem")) { if (ignmem == FALSE) { /* set */ if (set_mode == SET_LIMIT_SET) { retval = getsize(pres, &value); if (retval != PBSE_NONE) { sprintf(log_buffer, "getsize() failed for mem/pmem in %s", id); return(error(pname, retval)); } reslim.rlim_cur = reslim.rlim_max = value; #ifdef HAVE_GYGWIN_SETRLIMIT if (setrlimit(RLIMIT_DATA, &reslim) < 0) { sprintf(log_buffer, "cannot set data limit to %ld for job %s (setrlimit failed w/errno=%d (%s) - check default user limits)", (long int)reslim.rlim_max, pjob->ji_qs.ji_jobid, errno, strerror(errno)); return(error("RLIMIT_DATA", PBSE_SYSTEM)); } if (setrlimit(RLIMIT_RSS, &reslim) < 0) { sprintf(log_buffer, "cannot set RSS limit to %ld for job %s (setrlimit failed w/errno=%d (%s) - check default user limits)", (long int)reslim.rlim_max, pjob->ji_qs.ji_jobid, errno, strerror(errno)); return(error("RLIMIT_RSS", PBSE_SYSTEM)); } #endif /* HAVE_GYGWIN_SETRLIMIT */ mem_limit = value; if (getrlimit(RLIMIT_STACK, &reslim) >= 0) { /* NOTE: mem_limit no longer used with UMU patch in place */ mem_limit = value + reslim.rlim_cur; } } } } /* END else if (!strcmp(pname,"mem") && ... */ else if (!strcmp(pname, "walltime")) { /* check */ retval = gettime(pres, &value); if (retval != PBSE_NONE) { sprintf(log_buffer, "gettime() failed for walltime in %s\n", id); return(error(pname, retval)); } } else if (!strcmp(pname, "nice")) { /* set nice */ if (set_mode == SET_LIMIT_SET) { errno = 0; if ((nice((int)pres->rs_value.at_val.at_long) == -1) && (errno != 0)) { sprintf(log_buffer, "nice() failed w/errno=%d (%s) in %s\n", errno, strerror(errno), id); return(error(pname, PBSE_BADATVAL)); } } } else if (!strcmp(pname, "size")) { /* ignore */ /* NO-OP */ } else if(!strcmp(pname, "prologue")) { } else if(!strcmp(pname, "epilogue")) { } else if ((pres->rs_defin->rs_flags & ATR_DFLAG_RMOMIG) == 0) { /* don't recognize and not marked as ignore by mom */ sprintf(log_buffer, "do not know how to process resource '%s' in %s\n", pname, id); return(error(pname, PBSE_UNKRESC)); } pres = (resource *)GET_NEXT(pres->rs_link); } if (set_mode == SET_LIMIT_SET) { /* if either of vmem or pvmem was given, set sys limit to lesser */ if (vmem_limit != 0) { /* Don't make (p)vmem < pmem */ if (mem_limit > vmem_limit) { vmem_limit = mem_limit; } reslim.rlim_cur = reslim.rlim_max = vmem_limit; #ifdef HAVE_GYGWIN_SETRLIMIT if ((ignvmem == 0) && (setrlimit(RLIMIT_AS, &reslim) < 0)) { sprintf(log_buffer, "setrlimit() failed setting AS for vmem_limit mod in %s\n", id); return(error("RLIMIT_AS", PBSE_SYSTEM)); } #endif /* HAVE_GYGWIN_SETRLIMIT */ } } if (LOGLEVEL >= 5) { sprintf(log_buffer, "%s(%s,%s) completed", id, (pjob != NULL) ? pjob->ji_qs.ji_jobid : "NULL", (set_mode == SET_LIMIT_SET) ? "set" : "alter"); log_record(PBSEVENT_SYSTEM, 0, id, log_buffer); log_buffer[0] = '\0'; } /* SUCCESS */ return(PBSE_NONE); } /* END mom_set_limits() */ /* * State whether MOM main loop has to poll this job to determine if some * limits are being exceeded. * * Sets flag TRUE if polling is necessary, FALSE otherwise. Actual * polling is done using the mom_over_limit machine-dependent function. */ int mom_do_poll( job *pjob) /* I */ { char *pname; resource *pres; assert(pjob != NULL); if (LOGLEVEL >= 4) { log_record( PBSEVENT_JOB, PBS_EVENTCLASS_JOB, pjob->ji_qs.ji_jobid, "evaluating limits for job"); } assert(pjob != NULL); assert(pjob->ji_wattr[(int)JOB_ATR_resource].at_type == ATR_TYPE_RESC); pres = (resource *)GET_NEXT( pjob->ji_wattr[(int)JOB_ATR_resource].at_val.at_list); while (pres != NULL) { assert(pres->rs_defin != NULL); pname = pres->rs_defin->rs_name; assert(pname != NULL); assert(*pname != '\0'); if (strcmp(pname, "walltime") == 0 || strcmp(pname, "cput") == 0 || strcmp(pname, "pcput") == 0 || strcmp(pname, "pvmem") == 0 || strcmp(pname, "vmem") == 0) { return(TRUE); } pres = (resource *)GET_NEXT(pres->rs_link); } return(FALSE); } /* END mom_do_poll() */ /* * Setup for polling. * * Open kernel device and get namelist info. */ int mom_open_poll(void) { char *id = "mom_open_poll"; if (LOGLEVEL >= 6) { log_record(PBSEVENT_SYSTEM, 0, id, "started"); } pagesize = getpagesize(); proc_array = (proc_stat_t *)calloc(TBL_INC, sizeof(proc_stat_t)); if (proc_array == NULL) { log_err(errno, id, "malloc"); return(PBSE_SYSTEM); } max_proc = TBL_INC; return(PBSE_NONE); } /* END mom_open_poll() */ /* * Declare start of polling loop. * * This function caches information about all of processes * on the compute node (pbs_mom calls this function). Each process * in /proc/ is queried by looking at the 'stat' file. Statistics like * CPU usage time, memory consumption, etc. are gathered in the proc_array * list. This list is then used throughout the pbs_mom to get information * about tasks it is monitoring. * * This function is called from the main MOM loop once every "check_poll_interval" * seconds. * * @see get_proc_stat() - child * @see mom_set_use() - Aggregates data collected here * * NOTE: populates global 'proc_array[]' variable. * NOTE: reallocs proc_array[] as needed to accomodate processes. * * @see mom_open_poll() - allocs proc_array table. * @see mom_close_poll() - frees procs_array. * @see setup_program_environment() - parent - called at pbs_mom start * @see main_loop() - parent - called once per iteration * @see mom_set_use() - populate job structure with usage data for local use or to send to mother superior */ int mom_get_sample(void) { char *id = "mom_get_sample"; struct dirent *dent; proc_stat_t *pi; proc_stat_t *ps; rewinddir(pdir); nproc = 0; pi = proc_array; if (LOGLEVEL >= 6) { log_record(PBSEVENT_DEBUG, PBS_EVENTCLASS_SERVER, id, "proc_array load started"); } while ((dent = readdir(pdir)) != NULL) { if (!isdigit(dent->d_name[0])) continue; if ((ps = get_proc_stat(atoi(dent->d_name))) == NULL) { if (errno != ENOENT) { sprintf(log_buffer, "%s: get_proc_stat", dent->d_name); log_err(errno, id, log_buffer); } continue; } /* nproc++; -- we need to increment AFTER assigning this ps to the proc_array--otherwise we could skip it in for loops */ if ((nproc + 1) >= max_proc) { proc_stat_t *hold; if (LOGLEVEL >= 8) { log_record(PBSEVENT_DEBUG, PBS_EVENTCLASS_SERVER, id, "alloc more proc_array"); } max_proc *= 2; hold = (proc_stat_t *)realloc(proc_array, (max_proc + 1) * sizeof(proc_stat_t)); if (proc_array == NULL) { log_err(errno, id, "unable to realloc space for proc_array sample"); return(PBSE_SYSTEM); } proc_array = hold; memset(&proc_array[(nproc+1)], '\0', sizeof(proc_stat_t) * (max_proc >> 1)); } /* END if ((nproc+1) == max_proc) */ pi = &proc_array[nproc++]; memcpy(pi, ps, sizeof(proc_stat_t)); } /* END while ((dent = readdir(pdir)) != NULL) */ if (LOGLEVEL >= 6) { sprintf(log_buffer, "proc_array loaded - nproc=%d", nproc); log_record(PBSEVENT_DEBUG, 0, id, log_buffer); } return(PBSE_NONE); } /* END mom_get_sample() */ /* * Measure job resource usage and compare with its limits. * * If it has exceeded any well-formed polled limit return TRUE. * Otherwise, return FALSE. log_buffer is populated with failure. */ int mom_over_limit( job *pjob) /* I */ { char *pname; int retval; unsigned long value; unsigned long num; unsigned long long numll; resource *pres; assert(pjob != NULL); assert(pjob->ji_wattr[(int)JOB_ATR_resource].at_type == ATR_TYPE_RESC); pres = (resource *)GET_NEXT( pjob->ji_wattr[(int)JOB_ATR_resource].at_val.at_list); for (;pres != NULL;pres = (resource *)GET_NEXT(pres->rs_link)) { assert(pres->rs_defin != NULL); pname = pres->rs_defin->rs_name; assert(pname != NULL); assert(*pname != '\0'); if ((igncput == FALSE) && (strcmp(pname, "cput") == 0)) { retval = gettime(pres, &value); if (retval != PBSE_NONE) continue; if ((num = cput_sum(pjob)) > value) { sprintf(log_buffer, "cput %lu exceeded limit %lu", num, value); return(TRUE); } } else if ((igncput == FALSE) && (strcmp(pname, "pcput") == 0)) { retval = gettime(pres, &value); if (retval != PBSE_NONE) continue; if (overcpu_proc(pjob, value)) { sprintf(log_buffer, "pcput exceeded limit %lu", value); return(TRUE); } } else if (strcmp(pname, "vmem") == 0) { retval = getsize(pres, &value); if (retval != PBSE_NONE) continue; if ((ignvmem == 0) && ((numll = mem_sum(pjob)) > value)) { sprintf(log_buffer, "vmem %llu exceeded limit %lu", numll, value); return(TRUE); } } else if (strcmp(pname, "pvmem") == 0) { unsigned long long valuell; retval = getsize(pres, &value); if (retval != PBSE_NONE) continue; valuell = (unsigned long long)value; if ((ignvmem == 0) && (overmem_proc(pjob, valuell))) { sprintf(log_buffer, "pvmem exceeded limit %llu", valuell); return(TRUE); } } else if (ignwalltime == 0 && strcmp(pname, "walltime") == 0) { /* no need to check walltime on sisters, MS will get it */ if ((pjob->ji_qs.ji_svrflags & JOB_SVFLG_HERE) == 0) continue; retval = gettime(pres, &value); if (retval != PBSE_NONE) continue; num = (unsigned long)((double)(time_now - pjob->ji_qs.ji_stime) * wallfactor); if (num > value) { sprintf(log_buffer, "walltime %ld exceeded limit %ld", num, value); return(TRUE); } } } /* END for (pres) */ return(FALSE); } /* END mom_over_limit() */ /* * Update the job attribute for resources used. * * The first time this function is called for a job, * it sets up resource entries for * each resource that can be reported for this machine. * * Subsequent calls update the resource usage information based on * stats gathered by the mom_get_sample() function. This function * is often called by "im_request()" as a result of POLL_JOB query * from the mother superior. * * @see im_request() - parent - respond to poll_job request from mother superior * @see examine_all_running_jobs() - parent - update local use on mother superior * @see TMomFinalizeJob1() - parent - update serial job immediately at job start * * @return An error code if something goes wrong. */ int mom_set_use( job *pjob) /* I (modified) */ { resource *pres; attribute *at; resource_def *rd; unsigned long *lp, lnum; assert(pjob != NULL); at = &pjob->ji_wattr[(int)JOB_ATR_resc_used]; assert(at->at_type == ATR_TYPE_RESC); at->at_flags |= ATR_VFLAG_MODIFY; if ((at->at_flags & ATR_VFLAG_SET) == 0) { /* initialize usage structures */ at->at_flags |= ATR_VFLAG_SET; rd = find_resc_def(svr_resc_def, "cput", svr_resc_size); assert(rd != NULL); pres = add_resource_entry(at, rd); pres->rs_value.at_flags |= ATR_VFLAG_SET; pres->rs_value.at_type = ATR_TYPE_LONG; rd = find_resc_def(svr_resc_def, "vmem", svr_resc_size); assert(rd != NULL); pres = add_resource_entry(at, rd); pres->rs_value.at_flags |= ATR_VFLAG_SET; pres->rs_value.at_type = ATR_TYPE_SIZE; pres->rs_value.at_val.at_size.atsv_shift = 10; /* KB */ pres->rs_value.at_val.at_size.atsv_units = ATR_SV_BYTESZ; rd = find_resc_def(svr_resc_def, "walltime", svr_resc_size); assert(rd != NULL); pres = add_resource_entry(at, rd); pres->rs_value.at_flags |= ATR_VFLAG_SET; pres->rs_value.at_type = ATR_TYPE_LONG; rd = find_resc_def(svr_resc_def, "mem", svr_resc_size); assert(rd != NULL); pres = add_resource_entry(at, rd); pres->rs_value.at_flags |= ATR_VFLAG_SET; pres->rs_value.at_type = ATR_TYPE_SIZE; pres->rs_value.at_val.at_size.atsv_shift = 10; /* KB */ pres->rs_value.at_val.at_size.atsv_units = ATR_SV_BYTESZ; } /* END if ((at->at_flags & ATR_VFLAG_SET) == 0) */ /* get cputime */ rd = find_resc_def(svr_resc_def, "cput", svr_resc_size); assert(rd != NULL); pres = find_resc_entry(at, rd); assert(pres != NULL); lp = (unsigned long *) & pres->rs_value.at_val.at_long; lnum = cput_sum(pjob); *lp = MAX(*lp, lnum); /* get swap */ rd = find_resc_def(svr_resc_def, "vmem", svr_resc_size); assert(rd != NULL); pres = find_resc_entry(at, rd); assert(pres != NULL); lp = &pres->rs_value.at_val.at_size.atsv_num; lnum = (mem_sum(pjob) + 1023) >> pres->rs_value.at_val.at_size.atsv_shift; /* as KB */ *lp = MAX(*lp, lnum); /* get walltime */ rd = find_resc_def(svr_resc_def, "walltime", svr_resc_size); assert(rd != NULL); pres = find_resc_entry(at, rd); assert(pres != NULL); /* NOTE: starting jobs can come through here before stime is recorded */ if (pjob->ji_qs.ji_stime == 0) pres->rs_value.at_val.at_long = 0; else pres->rs_value.at_val.at_long = (long)((double)(time_now - pjob->ji_qs.ji_stime) * wallfactor); /* get memory */ rd = find_resc_def(svr_resc_def, "mem", svr_resc_size); assert(rd != NULL); pres = find_resc_entry(at, rd); assert(pres != NULL); lp = &pres->rs_value.at_val.at_size.atsv_num; lnum = (resi_sum(pjob) + 1023) >> pres->rs_value.at_val.at_size.atsv_shift; /* as KB */ *lp = MAX(*lp, lnum); return(PBSE_NONE); } /* END mom_set_use() */ /** * Kill a task session. * Call with the task pointer and a signal number. * * @return number of tasks signalled (0 = failure) * * @see kill_job() - parent * * NOTE: should support killpg() or killpidtree() - (NYI) * may be required for suspend/resume */ int kill_task( task *ptask, /* I */ int sig, /* I */ int pg) /* I (1=signal process group, 0=signal master process only) */ { char *id = "kill_task"; int ct = 0; /* num of processes killed */ int NumProcessesFound = 0; /* number of processes found with session ID */ struct dirent *dent; proc_stat_t *ps; int sesid; pid_t mompid; sesid = ptask->ti_qs.ti_sid; mompid = getpid(); if (LOGLEVEL >= 5) { sprintf(log_buffer, "%s: sending signal %d to task %d, session %d", id, sig, ptask->ti_qs.ti_task, sesid); log_record( PBSEVENT_JOB, PBS_EVENTCLASS_JOB, ptask->ti_job->ji_qs.ji_jobid, log_buffer); } if (sesid <= 1) { if (LOGLEVEL >= 3) { sprintf(log_buffer, "cannot send signal %d to task (no session id)", sig); log_record( PBSEVENT_ERROR, PBS_EVENTCLASS_JOB, ptask->ti_job->ji_qs.ji_jobid, log_buffer); } /* FAILURE */ return(0); } /* pdir is global */ /* NOTE: do not use cached proc-buffer since we need up-to-date info */ /* pdir is global */ rewinddir(pdir); while ((dent = readdir(pdir)) != NULL) { if (!isdigit(dent->d_name[0])) continue; if ((ps = get_proc_stat(atoi(dent->d_name))) == NULL) { if (errno != ENOENT) { sprintf(log_buffer, "%s: get_proc_stat", dent->d_name); log_err(errno,id,log_buffer); } continue; } if ((sesid == ps->session) || (ProcIsChild(procfs,dent->d_name,ptask->ti_job->ji_qs.ji_jobid) == TRUE)) { NumProcessesFound++; if ((ps->state == 'Z') || (ps->pid == 0)) { /* * Killing a zombie is sure death! Its pid is zero, * which to kill(2) means 'every process in the process * group of the current process'. */ sprintf(log_buffer, "%s: not killing process (pid=%d/state=%c) with sig %d", id, ps->pid, ps->state, sig); log_record( PBSEVENT_JOB, PBS_EVENTCLASS_JOB, ptask->ti_job->ji_qs.ji_jobid, log_buffer); } /* END if ((ps->state == 'Z') || (ps->pid == 0)) */ else { int i = 0; if (ps->pid == mompid) { /* * there is a race condition with newly started jobs that * can be killed before they've established their own * session id. This means the child tasks still have MOM's * session id. We check this to make sure MOM doesn't kill * herself. */ if (LOGLEVEL >= 3) { sprintf(log_buffer, "%s: not killing process %d. Avoid sending signal because child task still has MOM's session id", id, ps->pid); log_record( PBSEVENT_JOB, PBS_EVENTCLASS_JOB, ptask->ti_job->ji_qs.ji_jobid, log_buffer); } if (LOGLEVEL >= 3) { sprintf(log_buffer, "%s: not killing process %d. Avoid sending signal because child task still has MOM's session id", id, ps->pid); log_record( PBSEVENT_JOB, PBS_EVENTCLASS_JOB, ptask->ti_job->ji_qs.ji_jobid, log_buffer); } continue; } /* END if (ps->pid == mompid) */ if (sig == SIGKILL) { struct timespec req; req.tv_sec = 0; req.tv_nsec = 250000000; /* .25 seconds */ /* give the process some time to quit gracefully first (up to .25*20=5 seconds) */ sprintf(log_buffer, "%s: killing pid %d task %d gracefully with sig %d", id, ps->pid, ptask->ti_qs.ti_task, SIGTERM); log_record( PBSEVENT_JOB, PBS_EVENTCLASS_JOB, ptask->ti_job->ji_qs.ji_jobid, log_buffer); if (pg == 0) kill(ps->pid, SIGTERM); else killpg(ps->pid, SIGTERM); for (i = 0;i < 20;i++) { /* check if process is gone */ if (kill(ps->pid, 0) == -1) break; nanosleep(&req, NULL); } /* END for (i = 0) */ } /* END if (sig == SIGKILL) */ else { i = 20; } if (i >= 20) { /* NOTE: handle race-condition where process goes zombie as a result of previous SIGTERM */ /* update proc info from /proc//stat */ if ((ps = get_proc_stat(ps->pid)) != NULL) { if (ps->state == 'Z') { /* * Killing a zombie is sure death! Its pid is zero, * which to kill(2) means 'every process in the process * group of the current process'. */ sprintf(log_buffer, "%s: not killing process (pid=%d/state=%c) with sig %d", id, ps->pid, ps->state, sig); log_record( PBSEVENT_JOB, PBS_EVENTCLASS_JOB, ptask->ti_job->ji_qs.ji_jobid, log_buffer); } /* END if ((ps->state == 'Z') || (ps->pid == 0)) */ else { /* kill process hard */ /* why is this not killing with SIGKILL? */ sprintf(log_buffer, "%s: killing pid %d task %d with sig %d", id, ps->pid, ptask->ti_qs.ti_task, sig); log_record( PBSEVENT_JOB, PBS_EVENTCLASS_JOB, ptask->ti_job->ji_qs.ji_jobid, log_buffer); if (pg == 0) kill(ps->pid, sig); else killpg(ps->pid, sig); } } /* END if ((ps = get_proc_stat(ps->pid)) != NULL) */ } /* END if (i >= 20) */ ++ct; } /* END else ((ps->state == 'Z') || (ps->pid == 0)) */ } /* END if (sesid == ps->session) */ } /* END while ((dent = readdir(pdir)) != NULL) */ /* NOTE: to fix bad state situations resulting from a hard crash, the logic below should be triggered any time no processes are found (NYI) */ if (IS_ADOPTED_TASK(ptask->ti_qs.ti_task) && (NumProcessesFound == 0)) { /* no process was found, but for an adopted task this is OK (we don't find * out about the adopted task's termination via waitpid()--so we can safely * say that we have "killed" the task, even though the task was killed/died * some other way */ ct++; /* do code to mark task as finished (borrowed from Linux scan_for_terminating())... */ ptask->ti_qs.ti_exitstat = 0; /* assume successful completion */ ptask->ti_qs.ti_status = TI_STATE_EXITED; task_save(ptask); sprintf(log_buffer, "%s: job %s adopted task %d was marked as terminated because task's PID was no longer found, sid=%d", id, ptask->ti_job->ji_qs.ji_jobid, ptask->ti_qs.ti_task, ptask->ti_qs.ti_sid); LOG_EVENT( PBSEVENT_DEBUG, PBS_EVENTCLASS_JOB, ptask->ti_job->ji_qs.ji_jobid, log_buffer); } if ((NumProcessesFound == 0) && (ct <= 0)) { /* we can't find any processes belonging to given session, so we can safely say * that we "killed" the task and have TORQUE clean it up */ ct++; /* do code to mark task as finished (borrowed from Linux scan_for_terminating())... */ ptask->ti_qs.ti_exitstat = 0; /* assume successful completion */ ptask->ti_qs.ti_status = TI_STATE_EXITED; task_save(ptask); if (LOGLEVEL >= 5) { sprintf(log_buffer, "%s: could not send signal %d to task %d (session %d)--no process was found with this session ID (marking task as killed)!", id, sig, ptask->ti_qs.ti_task, sesid); log_record( PBSEVENT_JOB, PBS_EVENTCLASS_JOB, ptask->ti_job->ji_qs.ji_jobid, log_buffer); } } /* SUCCESS */ return(ct); } /* END kill_task() */ /* * Clean up everything related to polling. */ int mom_close_poll(void) { char *id = "mom_close_poll"; if (LOGLEVEL >= 6) { log_record( PBSEVENT_SYSTEM, 0, id, "entered"); } if (pdir != NULL) { if (closedir(pdir) != 0) { log_err(errno, id, "closedir"); return(PBSE_SYSTEM); } pdir = NULL; } if (proc_array != NULL) { free(proc_array); } return(PBSE_NONE); } /* END mom_close_poll() */ /* * mom_does_checkpoint * * @returns CST values as described in resmon.h. */ int mom_does_checkpoint(void) { return(CST_BLCR); /* Use the BLCR checkpointing system. */ } /* * Checkpoint the job. * * If abort is true, kill it too. */ int mach_checkpoint( task *ptask, /* I */ char *file, /* I */ int abort) /* I */ { return(-1); } /* END mach_checkpoint() */ /* * Restart the job from the checkpoint file. * * Return -1 on error or sid if okay. */ long mach_restart( task *ptask, char *file) { return(-1); } #define dsecs(val) ( (double)(val) ) char *cput_job( pid_t jobid) { char *id = "cput_job"; int found = 0; int i; double cputime, addtime; proc_stat_t *ps; cputime = 0.0; if (LOGLEVEL >= 6) { sprintf(log_buffer, "proc_array loop start - jobid = %d", jobid); log_record(PBSEVENT_DEBUG, 0, id, log_buffer); } for (i = 0;i < nproc;i++) { ps = &proc_array[i]; if (jobid != ps->session) continue; found = 1; /* add utime and stime (AKE) */ addtime = dsecs(ps->utime) + dsecs(ps->stime) + dsecs(ps->cutime) + dsecs(ps->cstime); cputime += addtime; DBPRT(("%s: total %.2f pid %d %.2f\n", id, cputime, ps->pid, addtime)) } /* END for (i) */ if (!found) { rm_errno = RM_ERR_EXIST; return(NULL); } sprintf(ret_string, "%.2f", cputime * cputfactor); return(ret_string); } /* END cput_job() */ char *cput_proc( pid_t pid) { char *id = "cput_proc"; double cputime; proc_stat_t *ps; cputime = 0.0; if ((ps = get_proc_stat(pid)) == NULL) { if (errno != ENOENT) { sprintf(log_buffer, "%d: get_proc_stat", pid); log_err(errno, id, log_buffer); } rm_errno = RM_ERR_SYSTEM; return(NULL); } cputime = dsecs(ps->utime) + dsecs(ps->stime); sprintf(ret_string, "%.2f", cputime * cputfactor); return(ret_string); } /* END cput_proc() */ char *cput( struct rm_attribute *attrib) { char *id = "cput"; int value; if (attrib == NULL) { log_err(-1, id, no_parm); rm_errno = RM_ERR_NOPARAM; return(NULL); } if ((value = atoi(attrib->a_value)) == 0) { sprintf(log_buffer, "bad param: %s", attrib->a_value); log_err(-1, id, log_buffer); rm_errno = RM_ERR_BADPARAM; return(NULL); } if (momgetattr(NULL)) { log_err(-1, id, extra_parm); rm_errno = RM_ERR_BADPARAM; return(NULL); } if (strcmp(attrib->a_qualifier, "session") == 0) { return(cput_job((pid_t)value)); } if (strcmp(attrib->a_qualifier, "proc") == 0) { return(cput_proc((pid_t)value)); } rm_errno = RM_ERR_BADPARAM; return(NULL); } /* END cput() */ char *mem_job( pid_t sid) /* I */ { static char id[] = "mem_job"; unsigned long long memsize; int i; proc_stat_t *ps; /* max memsize ??? */ memsize = 0; if (LOGLEVEL >= 6) { sprintf(log_buffer, "proc_array loop start - sid = %d", sid); log_record(PBSEVENT_DEBUG, 0, id, log_buffer); } for (i = 0;i < nproc;i++) { ps = &proc_array[i]; if (sid != ps->session) continue; memsize += ps->vsize; } /* END for (i) */ if (memsize == 0) { rm_errno = RM_ERR_EXIST; return(NULL); } sprintf(ret_string, "%llukb", memsize >> 10); /* KB */ return(ret_string); } /* END mem_job() */ char *mem_proc( pid_t pid) { char *id = "mem_proc"; proc_stat_t *ps; if ((ps = get_proc_stat(pid)) == NULL) { if (errno != ENOENT) { sprintf(log_buffer, "%d: get_proc_stat", pid); log_err(errno, id, log_buffer); } rm_errno = RM_ERR_SYSTEM; return(NULL); } sprintf(ret_string, "%llukb", (unsigned long long)ps->vsize >> 10); /* KB */ return(ret_string); } /* END mem_proc() */ char *mem( struct rm_attribute *attrib) { char *id = "mem"; int value; if (attrib == NULL) { log_err(-1, id, no_parm); rm_errno = RM_ERR_NOPARAM; return(NULL); } if ((value = atoi(attrib->a_value)) == 0) { sprintf(log_buffer, "bad param: %s", attrib->a_value); log_err(-1, id, log_buffer); rm_errno = RM_ERR_BADPARAM; return(NULL); } if (momgetattr(NULL)) { log_err(-1, id, extra_parm); rm_errno = RM_ERR_BADPARAM; return(NULL); } if (strcmp(attrib->a_qualifier, "session") == 0) { return(mem_job((pid_t)value)); } else if (strcmp(attrib->a_qualifier, "proc") == 0) { return(mem_proc((pid_t)value)); } else { rm_errno = RM_ERR_BADPARAM; return(NULL); } return(NULL); } /* END mem() */ static char *resi_job( pid_t jobid) { char *id = "resi_job"; ulong resisize; int found = 0; proc_stat_t *ps; int i; resisize = 0; if (LOGLEVEL >= 6) { sprintf(log_buffer, "proc_array loop start - jobid = %d", jobid); log_record(PBSEVENT_DEBUG, 0, id, log_buffer); } for (i = 0;i < nproc;i++) { ps = &proc_array[i]; if (jobid != ps->session) continue; found = 1; resisize += ps->rss; } /* END for (i) */ if (found) { /* in KB */ sprintf(ret_string, "%lukb", (resisize * (ulong)pagesize) >> 10); return(ret_string); } rm_errno = RM_ERR_EXIST; return(NULL); } /* END resi_job() */ static char *resi_proc( pid_t pid) { char *id = "resi_proc"; proc_stat_t *ps; if ((ps = get_proc_stat(pid)) == NULL) { if (errno != ENOENT) { sprintf(log_buffer, "%d: get_proc_stat(PIOCPSINFO)", pid); log_err(errno, id, log_buffer); } rm_errno = RM_ERR_SYSTEM; return(NULL); } /* in KB */ sprintf(ret_string, "%lukb", ((ulong)ps->rss * (ulong)pagesize) >> 10); return(ret_string); } /* END resi_proc() */ static char *resi( struct rm_attribute *attrib) { char *id = "resi"; int value; if (attrib == NULL) { log_err(-1, id, no_parm); rm_errno = RM_ERR_NOPARAM; return(NULL); } if ((value = atoi(attrib->a_value)) == 0) { sprintf(log_buffer, "bad param: %s", attrib->a_value); log_err(-1, id, log_buffer); rm_errno = RM_ERR_BADPARAM; return(NULL); } if (momgetattr(NULL)) { log_err(-1, id, extra_parm); rm_errno = RM_ERR_BADPARAM; return(NULL); } if (strcmp(attrib->a_qualifier, "session") == 0) { return(resi_job((pid_t)value)); } if (strcmp(attrib->a_qualifier, "proc") == 0) { return(resi_proc((pid_t)value)); } rm_errno = RM_ERR_BADPARAM; return(NULL); } /* END resi() */ char *sessions( struct rm_attribute *attrib) /* I */ { char *id = "sessions"; int i; int j; proc_stat_t *ps; char *fmt; int njids = 0; pid_t *jids, *hold; static int maxjid = 200; register pid_t jobid; if (attrib != NULL) { log_err(-1, id, extra_parm); rm_errno = RM_ERR_BADPARAM; return(NULL); } if ((jids = (pid_t *)calloc(maxjid, sizeof(pid_t))) == NULL) { log_err(errno, id, "no memory"); rm_errno = RM_ERR_SYSTEM; return(NULL); } /* Search for members of session */ for (i = 0;i < nproc;i++) { ps = &proc_array[i]; if (ps->uid == 0) continue; if ((jobid = ps->session) == 0) continue; if (LOGLEVEL >= 6) { sprintf(log_buffer, "%s[%d]: pid %d sid %d", id, njids, ps->pid, jobid); log_record( PBSEVENT_SYSTEM, 0, id, log_buffer); } for (j = 0;j < njids;j++) { if (jids[j] == jobid) break; } if (j == njids) { /* not found */ if (njids == maxjid) { /* need more space */ maxjid += 100; hold = (pid_t *)realloc(jids, maxjid); if (hold == NULL) { log_err(errno, id, "realloc"); rm_errno = RM_ERR_SYSTEM; free(jids); return(NULL); } jids = hold; } jids[njids++] = jobid; /* add jobid to list */ } } if (njids == 0) { free(jids); return(NULL); } fmt = ret_string; for (j = 0;j < njids;j++) { checkret(&fmt, 100); if (j == 0) { sprintf(fmt, "%d", (int)jids[j]); } else { sprintf(fmt, " %d", (int)jids[j]); } fmt += strlen(fmt); } /* END for (j) */ free(jids); return(ret_string); } /* END sessions() */ char *nsessions( struct rm_attribute *attrib) { char *result, *ch; int num = 1; if ((result = sessions(attrib)) == NULL) { return(result); } for (ch = result;*ch;ch++) { if (*ch == ' ') /* count blanks */ num++; } /* END for (ch) */ sprintf(ret_string, "%d", num); return(ret_string); } /* END nsessions() */ char *pids( struct rm_attribute *attrib) /* I */ { char *id = "pids"; pid_t jobid; proc_stat_t *ps; char *fmt; int i; int num_pids = 0; if (attrib == NULL) { log_err(-1, id, no_parm); rm_errno = RM_ERR_NOPARAM; return(NULL); } if ((jobid = (pid_t)atoi(attrib->a_value)) == 0) { sprintf(log_buffer, "bad param: %s", attrib->a_value); log_err(-1, id, log_buffer); rm_errno = RM_ERR_BADPARAM; return(NULL); } if (momgetattr(NULL)) { log_err(-1, id, extra_parm); rm_errno = RM_ERR_BADPARAM; return(NULL); } if (strcmp(attrib->a_qualifier, "session") != 0) { rm_errno = RM_ERR_BADPARAM; return(NULL); } /* Search for members of session */ fmt = ret_string; for (i = 0;i < nproc;i++) { ps = &proc_array[i]; if (LOGLEVEL >= 6) { DBPRT(("%s[%d]: pid: %d sid: %d\n", id, num_pids, ps->pid, ps->session)) } if (jobid != ps->session) continue; sprintf(fmt, "%d ", ps->pid); fmt += strlen(fmt); num_pids++; } /* END for (i) */ if (num_pids == 0) { rm_errno = RM_ERR_EXIST; return(NULL); } return(ret_string); } /* END pids() */ char *nusers( struct rm_attribute *attrib) { char *id = "nusers"; int j; int i; proc_stat_t *ps; int nuids = 0; uid_t *uids, *hold; static int maxuid = 200; register uid_t uid; if (attrib != NULL) { log_err(-1, id, extra_parm); rm_errno = RM_ERR_BADPARAM; return(NULL); } if ((uids = (uid_t *)calloc(maxuid, sizeof(uid_t))) == NULL) { log_err(errno, id, "no memory"); rm_errno = RM_ERR_SYSTEM; return(NULL); } for (i = 0;i < nproc;i++) { ps = &proc_array[i]; if ((uid = ps->uid) == 0) continue; if (LOGLEVEL >= 6) { sprintf(log_buffer, "%s[%d]: pid %d uid %d", id, nuids, ps->pid, uid); log_record(PBSEVENT_SYSTEM, 0, id, log_buffer); } for (j = 0;j < nuids;j++) { if (uids[j] == uid) break; } if (j == nuids) { /* not found */ if (nuids == maxuid) { /* need more space */ maxuid += 100; hold = (uid_t *)realloc(uids, maxuid); if (hold == NULL) { log_err(errno, id, "realloc"); rm_errno = RM_ERR_SYSTEM; free(uids); return(NULL); } uids = hold; } uids[nuids++] = uid; /* add uid to list */ } } /* END for (i) */ sprintf(ret_string, "%d", nuids); free(uids); return(ret_string); } /* END nusers() */ static char *totmem( struct rm_attribute *attrib) { char *id = "totmem"; proc_mem_t *mm; if (attrib) { log_err(-1, id, extra_parm); rm_errno = RM_ERR_BADPARAM; return(NULL); } if ((mm = get_proc_mem()) == NULL) { log_err(errno, id, "get_proc_mem"); rm_errno = RM_ERR_SYSTEM; return(NULL); } if (LOGLEVEL >= 6) { sprintf(log_buffer, "%s: total mem=%llu", id, mm->mem_total + mm->swap_total); log_record(PBSEVENT_SYSTEM, 0, id, log_buffer); } sprintf(ret_string, "%lukb", (ulong)((mm->mem_total >> 10) + (mm->swap_total >> 10))); /* KB */ return(ret_string); } /* END totmem() */ /* * Has been used Win API. * Cygwin's /proc/meminfo also takes information from here * but does it incorrectly till now */ static char *availmem( struct rm_attribute *attrib) { char *id = "availmem"; /* proc_mem_t *mm; */ MEMORYSTATUS memstat; if (attrib != NULL) { log_err(-1, id, extra_parm); rm_errno = RM_ERR_BADPARAM; return(NULL); } /* if ((mm = get_proc_mem()) == NULL) { log_err(errno, id, "get_proc_mem"); rm_errno = RM_ERR_SYSTEM; return(NULL); } */ GlobalMemoryStatus(&memstat); if (LOGLEVEL >= 6) { sprintf(log_buffer, "%s: free mem=%lu", id, memstat.dwAvailPageFile); log_record(PBSEVENT_SYSTEM, 0, id, log_buffer); } sprintf(ret_string, "%lukb", (ulong)(memstat.dwAvailPageFile >> 10)); /* KB */ return(ret_string); } /* END availmem() */ static char *ncpus( struct rm_attribute *attrib) { char *id = "ncpus"; char label[128]; FILE *fp; int procs; if (attrib != NULL) { log_err(-1, id, extra_parm); rm_errno = RM_ERR_BADPARAM; return(NULL); } if ((fp = fopen("/proc/cpuinfo", "r")) == NULL) { return(NULL); } procs = 0; while (!feof(fp)) { if (fscanf(fp, "%s %*[^\n]%*c", label) == 0) { getc(fp); /* must do something to get to eof */ } else if (strcmp("processor", label) == 0) procs++; } sprintf(ret_string, "%d", procs); system_ncpus = procs; fclose(fp); return(ret_string); } /* END ncpus() */ #define PMEMBUF_SIZE 2048 static const char path_meminfo[] = "/proc/meminfo"; static char *physmem( struct rm_attribute *attrib) { char *id = "physmem"; char tmpBuf[PMEMBUF_SIZE]; char *BPtr; int BSpace; unsigned long long mem; FILE *fp; if (attrib != NULL) { log_err(-1, id, extra_parm); rm_errno = RM_ERR_BADPARAM; return(NULL); } if (!(fp = fopen(path_meminfo, "r"))) { rm_errno = RM_ERR_SYSTEM; return(NULL); } BPtr = tmpBuf; BSpace = sizeof(tmpBuf); BPtr[0] = '\0'; while (!feof(fp)) { if (fgets(BPtr, BSpace, fp) == NULL) { break; } BSpace -= strlen(BPtr); BPtr += strlen(BPtr); } fclose(fp); /* FORMAT: '...\nMemTotal: XXX kB\n' */ if ((BPtr = strstr(tmpBuf, "MemTotal:")) != NULL) { BPtr += strlen("MemTotal:"); if (sscanf(BPtr, "%llu", &mem) != 1) { rm_errno = RM_ERR_SYSTEM; return(NULL); } /* value specified in kb */ } else { /* attempt to load first numeric value */ if (sscanf(BPtr, "%*s %llu", &mem) != 1) { rm_errno = RM_ERR_SYSTEM; return(NULL); } /* value specified in bytes */ mem >>= 10; } sprintf(ret_string, "%llukb", mem); return(ret_string); } /* END physmem() */ char *size_fs( char *param) { char *id = "size_fs"; struct statfs fsbuf; if (param[0] != '/') { sprintf(log_buffer, "%s: not full path filesystem name: %s", id, param); log_err(-1, id, log_buffer); rm_errno = RM_ERR_BADPARAM; return(NULL); } if (statfs(param, &fsbuf) == -1) { log_err(errno, id, "statfs"); rm_errno = RM_ERR_BADPARAM; return(NULL); } sprintf(ret_string, "%lukb:%lukb", (ulong)(((double)fsbuf.f_bsize * (double)fsbuf.f_bfree) / 1024.0), (ulong)(((double)fsbuf.f_bsize * (double)fsbuf.f_blocks) / 1024.0)); /* KB */ return(ret_string); } /* END size_fs() */ char *size_file( char *param) { char *id = "size_file"; struct stat sbuf; if (param[0] != '/') { sprintf(log_buffer, "%s: not full path filesystem name: %s", id, param); log_err(-1, id, log_buffer); rm_errno = RM_ERR_BADPARAM; return(NULL); } if (stat(param, &sbuf) == -1) { log_err(errno, id, "stat"); rm_errno = RM_ERR_BADPARAM; return(NULL); } sprintf(ret_string, "%lukb", (unsigned long)sbuf.st_size >> 10); /* KB */ return(ret_string); } /* END size_file() */ char *size( struct rm_attribute *attrib) { char *id = "size"; char *param; if (attrib == NULL) { log_err(-1, id, no_parm); rm_errno = RM_ERR_NOPARAM; return(NULL); } if (momgetattr(NULL)) { log_err(-1, id, extra_parm); rm_errno = RM_ERR_BADPARAM; return(NULL); } param = attrib->a_value; if (strcmp(attrib->a_qualifier, "file") == 0) { return(size_file(param)); } if (strcmp(attrib->a_qualifier, "fs") == 0) { return(size_fs(param)); } rm_errno = RM_ERR_BADPARAM; return(NULL); } /* END size() */ /* * For a recovering (-p) mom, look through existing tasks in existing * jobs for things that have exited that are not owned by us through a * parent-child relationship. Otherwise we cannot report back to tm * clients when tasks have exited. */ void scan_non_child_tasks(void) { char *id = "scan_non_child_tasks"; job *job; extern tlist_head svr_alljobs; DIR *pdir; /* use local pdir to prevent race conditions associated w/global pdir (VPAC) */ pdir = opendir("/proc"); for (job = GET_NEXT(svr_alljobs);job != NULL;job = GET_NEXT(job->ji_alljobs)) { task *task; for (task = GET_NEXT(job->ji_tasks);task != NULL;task = GET_NEXT(task->ti_jobtask)) { struct dirent *dent; int found; /* only check on tasks that we think should still be around */ if (task->ti_qs.ti_status != TI_STATE_RUNNING) continue; /* look for processes with this session id */ found = 0; /* NOTE: on linux systems, the session master should have pid == sessionid */ if (kill(task->ti_qs.ti_sid, 0) != -1) { found = 1; } else { /* session master cannot be found, look for other pid in session */ rewinddir(pdir); while ((dent = readdir(pdir)) != NULL) { proc_stat_t *ps; if (!isdigit(dent->d_name[0])) continue; ps = get_proc_stat(atoi(dent->d_name)); if (ps == NULL) continue; if (ps->session == task->ti_qs.ti_sid) { found = 1; break; } } /* END while ((dent) != NULL) */ } if (!found) { char buf[1024]; extern int exiting_tasks; sprintf(buf, "found exited session %d for task %d in job %s", task->ti_qs.ti_sid, task->ti_qs.ti_task, job->ji_qs.ji_jobid); log_event( PBSEVENT_JOB, PBS_EVENTCLASS_JOB, id, buf); task->ti_qs.ti_exitstat = 0; /* actually unknown */ task->ti_qs.ti_status = TI_STATE_EXITED; task_save(task); exiting_tasks = 1; } } } /* END for (job = GET_NEXT(svr_alljobs)) */ closedir(pdir); return; } /* END scan_non_child_tasks() */ /* * Cygwin has a virtual /dev/tty (and other /dev/...) only. * This path has a specified value of time of last access. * As variant, it is proposed to return time from a system reboot moment. */ char *idletime( struct rm_attribute *attrib) { char *id = "idletime"; if (attrib) { log_err(-1, id, extra_parm); rm_errno = RM_ERR_BADPARAM; return(NULL); } sprintf(ret_string, "%ld", GetTickCount()/1000); return(ret_string); } /* END idletime() */ static char *walltime( struct rm_attribute *attrib) { char *id = "walltime"; int value, job, found = 0; time_t now, start; proc_stat_t *ps; int i; if (attrib == NULL) { log_err(-1, id, no_parm); rm_errno = RM_ERR_NOPARAM; return(NULL); } if ((value = atoi(attrib->a_value)) == 0) { sprintf(log_buffer, "bad param: %s", attrib->a_value); log_err(-1, id, log_buffer); rm_errno = RM_ERR_BADPARAM; return(NULL); } if (momgetattr(NULL)) { log_err(-1, id, extra_parm); rm_errno = RM_ERR_BADPARAM; return(NULL); } if (strcmp(attrib->a_qualifier, "proc") == 0) { job = 0; } else if (strcmp(attrib->a_qualifier, "session") == 0) { job = 1; } else { rm_errno = RM_ERR_BADPARAM; return(NULL); } if ((now = time(NULL)) <= 0) { log_err(errno, id, "time"); rm_errno = RM_ERR_SYSTEM; return(NULL); } start = now; for (i = 0;i < nproc;i++) { ps = &proc_array[i]; if (job != 0) { if (value != ps->session) continue; } else { if (value != ps->pid) continue; } found = 1; start = MIN((unsigned)start, ps->start_time); } /* END for (i) */ if (found) { sprintf(ret_string, "%ld", (long)((double)(now - start) * wallfactor)); return(ret_string); } rm_errno = RM_ERR_EXIST; return(NULL); } /* END walltime() */ int get_la( double *rv) /* O */ { FILE *fp; char *id = "get_la"; float load; if ((fp = fopen("/proc/loadavg", "r")) == NULL) { rm_errno = RM_ERR_SYSTEM; return(rm_errno); } if (fscanf(fp, "%f", &load) != 1) { log_err(errno, id, "fscanf of load in /proc/loadavg"); fclose(fp); rm_errno = RM_ERR_SYSTEM; return(rm_errno); } *rv = (double)load; fclose(fp); return(0); } /* END get_la() */ u_long gracetime( u_long secs) { time_t now = time((time_t *)NULL); if (secs > (u_long)now) /* time is in the future */ return(secs - now); return(0); } /* May be it is required in the future */ static char *quota( struct rm_attribute *attrib) { char *id = "quota"; int type; dev_t dirdev; uid_t uid; struct stat sb; struct mntent *me; struct dqblk qi; FILE *m; struct passwd *pw; static char *type_array[] = { "harddata", "softdata", "currdata", "hardfile", "softfile", "currfile", "timedata", "timefile", NULL }; enum type_name { harddata, softdata, currdata, hardfile, softfile, currfile, timedata, timefile, type_end }; if (attrib == NULL) { log_err(-1, id, no_parm); rm_errno = RM_ERR_NOPARAM; return(NULL); } if (strcmp(attrib->a_qualifier, "type")) { sprintf(log_buffer, "unknown qualifier %s", attrib->a_qualifier); log_err(-1, id, log_buffer); rm_errno = RM_ERR_BADPARAM; return(NULL); } for (type = 0;type < type_end;type++) { if (strcmp(attrib->a_value, type_array[type]) == 0) break; } if (type == type_end) { /* check to see if command is legal */ sprintf(log_buffer, "bad param: %s=%s", attrib->a_qualifier, attrib->a_value); log_err(-1, id, log_buffer); rm_errno = RM_ERR_BADPARAM; return(NULL); } if ((attrib = momgetattr(NULL)) == NULL) { log_err(-1, id, no_parm); rm_errno = RM_ERR_NOPARAM; return(NULL); } if (strcmp(attrib->a_qualifier, "dir") != 0) { sprintf(log_buffer, "bad param: %s=%s", attrib->a_qualifier, attrib->a_value); log_err(-1, id, log_buffer); rm_errno = RM_ERR_BADPARAM; return(NULL); } if (attrib->a_value[0] != '/') /* must be absolute path */ { sprintf(log_buffer, "not an absolute path: %s", attrib->a_value); log_err(-1, id, log_buffer); rm_errno = RM_ERR_BADPARAM; return NULL; } if (stat(attrib->a_value, &sb) == -1) { sprintf(log_buffer, "stat: %s", attrib->a_value); log_err(errno, id, log_buffer); rm_errno = RM_ERR_EXIST; return NULL; } dirdev = (dev_t)sb.st_dev; DBPRT(("dir has devnum %d\n", (int)dirdev)) if ((m = setmntent(MOUNTED, "r")) == NULL) { log_err(errno, id, "setmntent"); rm_errno = RM_ERR_SYSTEM; return NULL; } while ((me = getmntent(m)) != NULL) { if (strcmp(me->mnt_type, MNTTYPE_IGNORE) == 0) continue; if (stat(me->mnt_dir, &sb) == -1) { sprintf(log_buffer, "stat: %s", me->mnt_dir); log_err(errno, id, log_buffer); continue; } if (LOGLEVEL >= 6) DBPRT(("%s\t%s\t%d\n", me->mnt_fsname, me->mnt_dir, (int)dirdev)) if (!memcmp(&sb.st_dev, &dirdev, sizeof(dev_t))) break; } endmntent(m); if (me == NULL) { sprintf(log_buffer, "filesystem %s not found", attrib->a_value); log_err(-1, id, log_buffer); rm_errno = RM_ERR_EXIST; return NULL; } #if defined(MNTOPT_NOQUOTA) if (hasmntopt(me, MNTOPT_NOQUOTA) != NULL) { sprintf(log_buffer, "no quotas on filesystem %s", me->mnt_dir); log_err(-1, id, log_buffer); rm_errno = RM_ERR_EXIST; return NULL; } #endif /* MNTOPT_NOQUOTA */ if ((attrib = momgetattr(NULL)) == NULL) { log_err(-1, id, no_parm); rm_errno = RM_ERR_NOPARAM; return NULL; } if (strcmp(attrib->a_qualifier, "user") != 0) { sprintf(log_buffer, "bad param: %s=%s", attrib->a_qualifier, attrib->a_value); log_err(-1, id, log_buffer); rm_errno = RM_ERR_BADPARAM; return NULL; } if ((uid = (uid_t)atoi(attrib->a_value)) == 0) { if ((pw = getpwnam_ext(attrib->a_value)) == NULL) { sprintf(log_buffer, "user not found: %s", attrib->a_value); log_err(-1, id, log_buffer); rm_errno = RM_ERR_EXIST; return NULL; } uid = pw->pw_uid; } /* Cygwin doesn't yet support quota * It is stub */ if (quotactl( QCMD(Q_GETQUOTA, USRQUOTA), me->mnt_fsname, uid, (caddr_t)&qi) == -1) { log_err(errno, id, "quotactl"); rm_errno = RM_ERR_SYSTEM; return(NULL); } /* sizes in KB */ switch (type) { case harddata: sprintf(ret_string, "%lukb", (u_long)qi.dqb_bhardlimit >> 10); break; case softdata: sprintf(ret_string, "%lukb", (u_long)qi.dqb_bsoftlimit >> 10); break; case currdata: #if defined(TENABLEQUOTA) sprintf(ret_string, "%lukb", (u_long)qi.dqb_curspace >> 10); #endif /* TENABLEQUOTA */ break; case hardfile: sprintf(ret_string, "%lu", (u_long)qi.dqb_ihardlimit); break; case softfile: sprintf(ret_string, "%lu", (u_long)qi.dqb_isoftlimit); break; case currfile: sprintf(ret_string, "%lu", (u_long)qi.dqb_curinodes); break; case timedata: sprintf(ret_string, "%lu", gracetime((u_long)qi.dqb_btime)); break; case timefile: sprintf(ret_string, "%lu", gracetime((u_long)qi.dqb_itime)); break; } /* END switch() */ return(ret_string); } /* END quota() */ /* needs gcc -liphlpapi */ static char *netload( struct rm_attribute *attrib) { MIB_IPSTATS Stats; PMIB_IPSTATS pStats=&Stats; if (NO_ERROR == GetIpStatistics(pStats)) { sprintf(ret_string, "%lu", pStats->dwInReceives + pStats->dwOutRequests); return(ret_string); } rm_errno = RM_ERR_SYSTEM; return (NULL); } /* END netload() */ mbool_t ProcIsChild( char *Dir, /* I */ char *PID, /* I */ char *JobID) /* I */ { return(FALSE); } /* END ProcIsChild() */ /* END mom_mach.c */