/* * Copyright (C) 1994-2021 Altair Engineering, Inc. * For more information, contact Altair at www.altair.com. * * This file is part of both the OpenPBS software ("OpenPBS") * and the PBS Professional ("PBS Pro") software. * * Open Source License Information: * * OpenPBS is free software. You can redistribute it and/or modify it under * the terms of the GNU Affero General Public License as published by the * Free Software Foundation, either version 3 of the License, or (at your * option) any later version. * * OpenPBS is distributed in the hope that it will be useful, but WITHOUT * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or * FITNESS FOR A PARTICULAR PURPOSE. See the GNU Affero General Public * License for more details. * * You should have received a copy of the GNU Affero General Public License * along with this program. If not, see . * * Commercial License Information: * * PBS Pro is commercially licensed software that shares a common core with * the OpenPBS software. For a copy of the commercial license terms and * conditions, go to: (http://www.pbspro.com/agreement.html) or contact the * Altair Legal Department. * * Altair's dual-license business model allows companies, individuals, and * organizations to create proprietary derivative works of OpenPBS and * distribute them - whether embedded or bundled with other software - * under a commercial license agreement. * * Use of Altair's trademarks, including but not limited to "PBS™", * "OpenPBS®", "PBS Professional®", and "PBS Pro™" and Altair's logos is * subject to Altair's trademark licensing policies. */ #include /* the master config generated by configure */ /** * @file */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "libpbs.h" #include "list_link.h" #include "log.h" #include "server_limits.h" #include "attribute.h" #include "resource.h" #include "job.h" #include "pbs_nodes.h" #include "mom_mach.h" #include "mom_func.h" #ifdef PMIX #include "mom_pmix.h" #endif #include "resmon.h" #include "mom_vnode.h" #include "libutil.h" #include "work_task.h" /** * @struct * struct release_info is used to parse the release information * of ProPack and linux distributions (i.e RHEL/SLES). * * Keep the sgi-release file information at index 0 * To load the ProPack information, index 0 is used. * To load the os information, index 1 thru end is used * to search which file is available. */ static struct release_info { char *file; char *pfx; char *srch; char *sep; } release_info[] = { {"/etc/redhat-release", "RHEL", "release", " "}, {"/etc/SuSE-release", "SLES", "VERSION", "="}, {"/etc/os-release", "SLES", "VERSION", "="}}; /** * @struct * struct libjob_support is used to hold verified and tested list of * , , . */ static struct libjob_support { char *osver; char *arch; char *libjobver; } libjob_support[] = { {"SLES10", "x86_64", "libjob.so"}, {"SLES11", "x86_64", "libjob.so"}, {"SLES12", "x86_64", "libjob.so.0"}, {"SLES12", "aarch64", "libjob.so.0"}, {"SLES15", "aarch64", "libjob.so.0"}, {"SLES15", "x86_64", "libjob.so.0"}}; /* Global Variables */ extern int exiting_tasks; extern char mom_host[]; extern pbs_list_head svr_alljobs; extern int termin_child; extern int num_acpus; extern int num_pcpus; extern int svr_delay_entry; extern pbs_list_head task_list_event; #if MOM_ALPS extern char *path_jobs; char *get_versioned_libname(); int find_in_lib(void *handle, char *plnam, char *psnam, void **psp); /** * This is a temporary kludge - this work should really be done by * pbs_sched: if the job is getting exclusive use of a vnode, we * will assign all the CPU * resources of the vnode to the created CPU set. Exclusive use of * a vnode is defined by a table in (currently) section E16.4 of the * GRUNT 2 document, q.v.. It is reproduced here * * Resource_List.place value * vnode "sharing" * value unset contains "share" contains "excl" * ---------------------------------------------------| * unset | share | share | excl | * |--------------|---------------|-------------------| * "default_shared" | share | share | excl | * |--------------|---------------|-------------------| * "default_excl" | excl | share | excl | * |--------------|---------------|-------------------| * "ignore_excl" | share | share | share | * |--------------|---------------|-------------------| * "force_excl" | excl | excl | excl | * |--------------------------------------------------- * * and reflected in the vnss[][] array below. * * This applies to ALPS because the Cray reservation has an EXCLUSIVE * or SHARED mode that is set from this table. */ enum vnode_sharing_state vnss[][rlplace_excl - rlplace_unset + 1] = { {isshared, isshared, isexcl}, /* VNS_UNSET */ {isshared, isshared, isexcl}, /* VNS_DFLT_SHARED */ {isexcl, isshared, isexcl}, /* VNS_DFLT_EXCL */ {isshared, isshared, isshared}, /* VNS_IGNORE_EXCL */ {isexcl, isexcl, isexcl}, /* VNS_FORCE_EXCL */ {isexcl, isshared, isexcl}, /* VNS_DFLT_EXCLHOST */ {isexcl, isexcl, isexcl} /* VNS_FORCE_EXCLHOST */ }; #ifndef MAX #define MAX(a, b) (((a) > (b)) ? (a) : (b)) #endif /** * @brief * getplacesharing sharing value for job place * * Compare the "place" string for a job with "excl" and "share" and * return the corresponding rlplace_value. * * @param[in] pjob the job of interest * * @return enum rlplace_value * * @par Side-effects * A log message is printed at DEBUG level. * * @par * This code was put in an externally * available function for use by the Cray project. * */ enum rlplace_value getplacesharing(job *pjob) { static resource_def *prsdef = NULL; enum rlplace_value rpv = rlplace_unset; resource *pplace; /* * Compute the "Resource_List.place" index for vnss[][]: */ prsdef = &svr_resc_def[RESC_PLACE]; if (prsdef != NULL) { char *placeval = NULL; pplace = find_resc_entry(get_jattr(pjob, JOB_ATR_resource), prsdef); if (pplace) placeval = pplace->rs_value.at_val.at_str; if (placeval != NULL) { if (place_sharing_check(placeval, PLACE_Excl)) rpv = rlplace_excl; else if (place_sharing_check(placeval, PLACE_ExclHost)) rpv = rlplace_excl; else if (place_sharing_check(placeval, PLACE_Shared)) rpv = rlplace_share; sprintf(log_buffer, "Resource_List.place = %s", placeval); log_event(PBSEVENT_DEBUG, PBS_EVENTCLASS_JOB, LOG_DEBUG, pjob->ji_qs.ji_jobid, log_buffer); } } return rpv; } /* These globals are initialized in ck_acct_facility_present. * * At a later time it may be better to relocate them to the machine * independent portion of the mom code if they find use by more * than a single machine/OS type */ int job_facility_present; int job_facility_enabled; int acct_facility_present; int acct_facility_active; jid_t (*jc_create)(); jid_t (*jc_getjid)(); void ck_acct_facility_present(void) { int ret1; int ret2; char *libjob; static void *handle1 = NULL; struct config *cptr; extern struct config *config_array; /* use of job_create defaults to True */ job_facility_enabled = 1; for (cptr = config_array; cptr != NULL; cptr++) { if (cptr->c_name == NULL || *cptr->c_name == 0) break; else if (strcasecmp(cptr->c_name, "pbs_jobcreate_workload_mgmt") == 0) { (void) set_boolean(__func__, cptr->c_u.c_value, &job_facility_enabled); } } /* multiple calls to dlopen with the same arguments do not cause multiple * copies of the library to get loaded into the proesses memory, they just * bump a reference count and return the same handle value. * If dlclose is issued when the reference count is 1, the library will be * unloaded from memory and any previous pointers obtained through calls to * dlsym will not be valid. */ job_facility_present = 0; acct_facility_present = 0; acct_facility_active = 0; /* * If job facility is turned off, don't call dlopen for job_create. */ if (job_facility_enabled == 0) goto done; libjob = get_versioned_libname(); if (libjob == NULL) { sprintf(log_buffer, "Could not find a supported job shared object"); log_event(PBSEVENT_ADMIN, PBS_EVENTCLASS_ACCT, LOG_DEBUG, __func__, log_buffer); goto err; } sprintf(log_buffer, "using %s for job shared object", libjob); log_event(PBSEVENT_ADMIN, PBS_EVENTCLASS_ACCT, LOG_DEBUG, __func__, log_buffer); handle1 = dlopen(libjob, RTLD_LAZY); if (handle1 == NULL) { /* facility is not available */ sprintf(log_buffer, "%s. failed to dlopen %s", dlerror(), libjob); log_event(PBSEVENT_ADMIN, PBS_EVENTCLASS_ACCT, LOG_DEBUG, __func__, log_buffer); goto err; } sprintf(log_buffer, "dlopen of %s successful", libjob); log_event(PBSEVENT_ADMIN, PBS_EVENTCLASS_ACCT, LOG_DEBUG, __func__, log_buffer); /* find_in_lib sets message in log_buffer */ ret1 = find_in_lib(handle1, libjob, "job_create", (void **) &jc_create); log_event(PBSEVENT_ADMIN, PBS_EVENTCLASS_ACCT, LOG_DEBUG, __func__, log_buffer); ret2 = find_in_lib(handle1, libjob, "job_getjid", (void **) &jc_getjid); log_event(PBSEVENT_ADMIN, PBS_EVENTCLASS_ACCT, LOG_DEBUG, __func__, log_buffer); if ((ret1 == 1) && (ret2 == 1)) job_facility_present = 1; if (job_facility_present == 0) goto done; err: log_event(PBSEVENT_ADMIN, PBS_EVENTCLASS_ACCT, LOG_DEBUG, __func__, "job facility not present or improperly setup"); done: /* * When we get here, the flags are set to indicate what libs should * be kept open. */ if (job_facility_present == 0) { if (handle1) { dlclose(handle1); handle1 = NULL; } } } /** * @brief * find_in_lib - Call this function when you want to find the address of symbol * in a shared library that has been opened by a call to dlopen. * * An appropriate message will be written into PBS' global "log_buffer" * in each of the three possible cases (found, not found, bogus arguments). * The caller chooses to log or ignore the content of log_buffer. * * * @param[in] handle valid handle from call to dlopen * @param[in] plnam pointer to the name of the library (NULL acceptable) * @param[in] psnam pointer to the name of the symbol * @param[out] psp where to return the symbol pointer if found * * @return int * @retval 1 success, with symbol pointer stored to *psp * @retval 0 failure, and *psp unmodified * @retval -1 bad input to this function * */ int find_in_lib(void *handle, char *plnam, char *psnam, void **psp) { void *psym; const char *error; int retcode; /* check arguments */ if (handle == NULL || psnam == NULL || *psnam == '\0') { sprintf(log_buffer, "%s: bad arguments %p %p %p %p", __func__, handle, plnam, psnam, psp); return -1; } psym = dlsym(handle, psnam); error = dlerror(); if (error != NULL) { retcode = 0; if (plnam) sprintf(log_buffer, "%s. symbol %s not found in %s", error, psnam, plnam); else sprintf(log_buffer, "%s. symbol %s not found", error, psnam); } else { retcode = 1; *psp = psym; if (plnam) sprintf(log_buffer, "symbol %s found in %s", psnam, plnam); else sprintf(log_buffer, "symbol %s found", psnam); } return (retcode); } #endif /* MOM_ALPS */ /* Private variables */ /** * @brief * Set session id and whatever else is required on this machine * to create a new job. * On a Cray, an ALPS reservation will be created and confirmed. * * @param[in] pjob - pointer to job structure * @param[in] sjr - pointer to startjob_rtn structure * * @return session/job id * @retval -1 error from setsid(), no message in log_buffer * @retval -2 temporary error, retry job, message in log_buffer * @retval -3 permanent error, abort job, message in log_buffer * */ int set_job(job *pjob, struct startjob_rtn *sjr) { #if MOM_ALPS if (job_facility_present && pjob->ji_qs.ji_svrflags & JOB_SVFLG_HERE) { /* host system has necessary JOB container facility present * and this host is Mother Superior for this job */ jid_t *pjid = (jid_t *) &pjob->ji_extended.ji_ext.ji_jid[0]; if (*pjid != (jid_t) 0 && *pjid != (jid_t) -1) { sjr->sj_jid = *pjid; } else { errno = -1; sjr->sj_jid = (jc_create == NULL) ? -1 : (*jc_create)(0, pjob->ji_qs.ji_un.ji_momt.ji_exuid, 0); if (sjr->sj_jid == (jid_t) -1) { /* Failed: categorize errno into two cases and handle */ /* Remark: sit_job call occurs before log_close() */ if (errno == ENOSYS) { if (job_facility_present == 1) { log_joberr(errno, __func__, "Job container facility unavailable", pjob->ji_qs.ji_jobid); job_facility_present = 0; } } else { /* log any other job_create failure type */ log_joberr(errno, __func__, "Job container job_create call failed", pjob->ji_qs.ji_jobid); } } } *pjid = sjr->sj_jid; } #endif /* MOM_ALPS */ sjr->sj_session = setsid(); #if MOM_ALPS /* * Now that we have our SID/JID we can request/confirm our * placement scheduler reservation. * * Do this only if we are mother superior for the job. */ if (pjob->ji_qs.ji_svrflags & JOB_SVFLG_HERE) { basil_request_reserve_t *basil_req; int rc; /* initialized to -1 so this catches the unset case. */ sjr->sj_reservation = -1; rc = alps_create_reserve_request(pjob, &basil_req); if (rc == 1) { sprintf(log_buffer, "Fatal MPP reservation error" " preparing request."); return -3; } else if (rc == 2) { sprintf(log_buffer, "Transient MPP reservation error" " preparing request."); return -2; } if (basil_req) { rc = alps_create_reservation(basil_req, &sjr->sj_reservation, &sjr->sj_pagg); alps_free_reserve_request(basil_req); if (rc < 0) { sprintf(log_buffer, "Fatal MPP reservation error" " on create."); return -3; } if (rc > 0) { sprintf(log_buffer, "Transient MPP reservation error" " on create."); return -2; } /* * If we are interacting with ALPS, the cookie has * not been set. Fill in the session ID we just * acquired. Otherwise, we are interacting with * CPA and use the cookie that was acquired when * the reservation was created. */ if (sjr->sj_pagg == 0) { if ((job_facility_present == 1)) sjr->sj_pagg = sjr->sj_jid; else sjr->sj_pagg = sjr->sj_session; } pjob->ji_extended.ji_ext.ji_reservation = sjr->sj_reservation; pjob->ji_extended.ji_ext.ji_pagg = sjr->sj_pagg; rc = alps_confirm_reservation(pjob); if (rc < 0) { sprintf(log_buffer, "Fatal MPP reservation error" " on confirm."); return -3; } if (rc > 0) { sprintf(log_buffer, "Transient MPP reservation error" " on confirm."); return -2; } } else { /* No error but no reservation made, reset so * the inventory will not be reread. */ sjr->sj_reservation = 0; } } #endif /* MOM_ALPS */ return (sjr->sj_session); } /** * @brief * set_globid - set the global id for a machine type. * * @param[in] pjob - pointer to job structure * @param[in] sjr - pointer to startjob_rtn structure * * @return Void * */ void set_globid(job *pjob, struct startjob_rtn *sjr) { #if MOM_ALPS char buf[19]; /* 0x,16 hex digits,'\0' */ char altid_buf[23]; if (sjr->sj_jid == (jid_t) -1) job_facility_present = 0; else if (sjr->sj_jid) { sprintf(buf, "%#0lx", (unsigned long) sjr->sj_jid); set_jattr_str_slim(pjob, JOB_ATR_acct_id, buf, NULL); (void) memcpy(&pjob->ji_extended.ji_ext.ji_jid, &sjr->sj_jid, sizeof(pjob->ji_extended.ji_ext.ji_jid)); if (job_facility_present == 0) { /* first success on job_create() after failure */ job_facility_present = 1; sprintf(log_buffer, "Job container facility available"); log_event(PBSEVENT_ADMIN, PBS_EVENTCLASS_ACCT, LOG_DEBUG, __func__, log_buffer); } } pjob->ji_extended.ji_ext.ji_pagg = sjr->sj_pagg; pjob->ji_extended.ji_ext.ji_reservation = sjr->sj_reservation; sprintf(altid_buf, "%ld", sjr->sj_reservation); set_jattr_str_slim(pjob, JOB_ATR_altid, altid_buf, NULL); #endif /* MOM_ALPS */ } /** * @brief * sets the shell to be used * * @param[in] pjob - pointer to job structure * @param[in] pwdp - pointer to passwd structure * * @return string * @retval shellname Success * */ char * set_shell(job *pjob, struct passwd *pwdp) { char *cp; int i; char *shell; struct array_strings *vstrs; /* * find which shell to use, one specified or the login shell */ shell = pwdp->pw_shell; if ((is_jattr_set(pjob, JOB_ATR_shell)) && (vstrs = get_jattr_arst(pjob, JOB_ATR_shell))) { for (i = 0; i < vstrs->as_usedptr; ++i) { cp = strchr(vstrs->as_string[i], '@'); if (cp) { if (!strncmp(mom_host, cp + 1, strlen(cp + 1))) { *cp = '\0'; /* host name matches */ shell = vstrs->as_string[i]; break; } } else { shell = vstrs->as_string[i]; /* wildcard */ } } } return (shell); } /** * * @brief * Checks if a child of the current (mom) process has terminated, and * matches it with the pid of one of the tasks in the task_list_event, * or matches the pid of a process being monitored for a PBS job. * if matching a task in the task_list_event, then that task is * marked as WORK_Deferred_Cmp along with the exit value of the child * process. Otherwise if it's for a job, and that job's * JOB_SVFLAG_TERMJOB is set, then mark the job as exiting. * * @return Void * */ void scan_for_terminated(void) { int exiteval; pid_t pid; job *pjob; task *ptask = NULL; struct work_task *wtask = NULL; int statloc; /* update the latest intelligence about the running jobs; */ /* must be done before we reap the zombies, else we lose the info */ termin_child = 0; mom_set_use_all(); /* Now figure out which task(s) have terminated (are zombies) */ while ((pid = waitpid(-1, &statloc, WNOHANG)) > 0) { if (WIFEXITED(statloc)) exiteval = WEXITSTATUS(statloc); else if (WIFSIGNALED(statloc)) exiteval = WTERMSIG(statloc) + 0x100; else exiteval = 1; /* Check for other task lists */ wtask = (struct work_task *) GET_NEXT(task_list_event); while (wtask) { if ((wtask->wt_type == WORK_Deferred_Child) && (wtask->wt_event == pid)) { wtask->wt_type = WORK_Deferred_Cmp; wtask->wt_aux = (int) exiteval; /* exit status */ svr_delay_entry++; /* see next_task() */ } wtask = (struct work_task *) GET_NEXT(wtask->wt_linkevent); } pjob = (job *) GET_NEXT(svr_alljobs); while (pjob) { /* ** see if process was a child doing a special ** function for MOM */ if (pid == pjob->ji_momsubt) break; /* ** look for task */ ptask = (task *) GET_NEXT(pjob->ji_tasks); while (ptask) { if (ptask->ti_qs.ti_sid == pid) break; ptask = (task *) GET_NEXT(ptask->ti_jobtask); } if (ptask != NULL) break; pjob = (job *) GET_NEXT(pjob->ji_alljobs); } if (pjob == NULL) { DBPRT(("%s: pid %d not tracked, exit %d\n", __func__, pid, exiteval)) continue; } if (pid == pjob->ji_momsubt) { pjob->ji_momsubt = 0; if (pjob->ji_mompost) { pjob->ji_mompost(pjob, exiteval); } (void) job_save(pjob); continue; } DBPRT(("%s: task %8.8X pid %d exit value %d\n", __func__, ptask->ti_qs.ti_task, pid, exiteval)) ptask->ti_qs.ti_exitstat = exiteval; sprintf(log_buffer, "task %8.8X terminated", ptask->ti_qs.ti_task); log_event(PBSEVENT_DEBUG, PBS_EVENTCLASS_JOB, LOG_DEBUG, pjob->ji_qs.ji_jobid, log_buffer); #ifdef PMIX /* Inform PMIx that the task has exited. */ pbs_pmix_notify_exit(pjob, ptask->ti_qs.ti_exitstat, NULL); #endif /* ** After the top process(shell) of the TASK exits, check if the ** JOB_SVFLG_TERMJOB job flag set. If yes, then check for any ** live process(s) in the session. If found, make the task ** ORPHAN by setting the flag and delay by kill_delay time. This ** will be exited in kill_job or by cput_sum() as can not be ** seen again by scan_for_terminated(). */ if (pjob->ji_qs.ji_svrflags & JOB_SVFLG_TERMJOB) { int n; (void) mom_get_sample(); n = bld_ptree(ptask->ti_qs.ti_sid); if (n > 0) { ptask->ti_flags |= TI_FLAGS_ORPHAN; DBPRT(("%s: task %8.8X still has %d active procs\n", __func__, ptask->ti_qs.ti_task, n)) continue; } } kill_session(ptask->ti_qs.ti_sid, SIGKILL, 0); ptask->ti_qs.ti_status = TI_STATE_EXITED; (void) task_save(ptask); exiting_tasks = 1; } } #ifdef HAVE_POSIX_OPENPT /** * @brief * This is code adapted from an example for posix_openpt in * The Open Group Base Specifications Issue 6. * * On success, this function returns an open descriptor for the * master pseudotty and places a pointer to the (static) name of * the slave pseudotty in *rtn_name; on failure, -1 is returned. * * @param[out] rtn_name - holds info of tty * * @return int * @retval fd Success * @retval -1 Failure * */ int open_master(char **rtn_name) { int masterfd; char *newslavename; static char slavename[_POSIX_PATH_MAX]; #ifndef _XOPEN_SOURCE extern char *ptsname(int); extern int grantpt(int); extern int unlockpt(int); extern int posix_openpt(int); #endif masterfd = posix_openpt(O_RDWR | O_NOCTTY); if (masterfd == -1) return (-1); if ((grantpt(masterfd) == -1) || (unlockpt(masterfd) == -1) || ((newslavename = ptsname(masterfd)) == NULL)) { (void) close(masterfd); return (-1); } pbs_strncpy(slavename, newslavename, sizeof(slavename)); assert(rtn_name != NULL); *rtn_name = slavename; return (masterfd); } #else /* HAVE_POSIX_OPENPT */ /** * @brief * creat the master pty, this particular * piece of code depends on multiplexor /dev/ptc * * @param[in] rtn_name - holds info about tty * @return int * @retval fd Success * @retval -1 Failure * */ #define PTY_SIZE 12 int open_master(char **rtn_name) { char *pc1; char *pc2; int ptc; /* master file descriptor */ static char ptcchar1[] = "pqrs"; static char ptcchar2[] = "0123456789abcdef"; static char pty_name[PTY_SIZE + 1]; /* "/dev/[pt]tyXY" */ pbs_strncpy(pty_name, "/dev/ptyXY", sizeof(pty_name)); for (pc1 = ptcchar1; *pc1 != '\0'; ++pc1) { pty_name[8] = *pc1; for (pc2 = ptcchar2; *pc2 != '\0'; ++pc2) { pty_name[9] = *pc2; if ((ptc = open(pty_name, O_RDWR | O_NOCTTY, 0)) >= 0) { /* Got a master, fix name to matching slave */ pty_name[5] = 't'; *rtn_name = pty_name; return (ptc); } else if (errno == ENOENT) return (-1); /* tried all entries, give up */ } } return (-1); /* tried all entries, give up */ } #endif /* HAVE_POSIX_OPENPT */ /* * struct sig_tbl = map of signal names to numbers, * see req_signal() in ../requests.c */ struct sig_tbl sig_tbl[] = { {"NULL", 0}, {"HUP", SIGHUP}, {"INT", SIGINT}, {"QUIT", SIGQUIT}, {"ILL", SIGILL}, {"TRAP", SIGTRAP}, {"IOT", SIGIOT}, {"ABRT", SIGABRT}, {"FPE", SIGFPE}, {"KILL", SIGKILL}, {"BUS", SIGBUS}, {"SEGV", SIGSEGV}, {"PIPE", SIGPIPE}, {"ALRM", SIGALRM}, {"TERM", SIGTERM}, {"URG", SIGURG}, {"STOP", SIGSTOP}, {"TSTP", SIGTSTP}, {"CONT", SIGCONT}, {"CHLD", SIGCHLD}, {"CLD", SIGCHLD}, {"TTIN", SIGTTIN}, {"TTOU", SIGTTOU}, {"IO", SIGIO}, #ifdef __linux__ {"POLL", SIGPOLL}, #endif {"XCPU", SIGXCPU}, {"XFSZ", SIGXFSZ}, {"VTALRM", SIGVTALRM}, {"PROF", SIGPROF}, {"WINCH", SIGWINCH}, {"USR1", SIGUSR1}, {"USR2", SIGUSR2}, {NULL, -1}}; /** * @brief * Get the release information * * @par Functionality: * This function extracts the release information of ProPack and Linux distributions * from system files listed in struct release_info. * * @see * get_versioned_lib * * @param[in] file - pointer to file * @param[in] pfx - pointer to prefix * @param[in] srch - pointer to search string * @param[in] sep - pointer to separator * * @return char * * @retval distro: or * @retval NULL: Not able to get the requested information from distro * * @par Side Effects: The value returned needs to be freed by the caller. * * @par MT-safe: Yes * */ static char * parse_sysfile_info(const char *file, const char *pfx, const char *srch, const char *sep) { FILE *fptr; char rbuf[1024]; char *tok; char *svptr = NULL; char *distro; int found = 0; fptr = fopen(file, "r"); if (fptr == NULL) return NULL; while (fgets(rbuf, sizeof(rbuf), fptr) != NULL) { if (strstr(rbuf, srch)) { found = 1; break; } } fclose(fptr); if (found == 0) { sprintf(log_buffer, "release info not found in %s", file); log_err(errno, __func__, log_buffer); return NULL; } tok = string_token(rbuf, sep, &svptr); while (tok) { if (strstr(tok, srch)) { tok = string_token(NULL, sep, &svptr); break; } tok = string_token(NULL, sep, &svptr); } if (tok == NULL) return NULL; while (!isdigit((int) (*tok))) tok++; distro = malloc(MAXNAMLEN); if (distro == NULL) { sprintf(log_buffer, "memory allocation failed"); log_err(errno, __func__, log_buffer); return NULL; } (void) snprintf(distro, MAXNAMLEN, "%s%d", pfx, atoi(tok)); distro[MAXNAMLEN - 1] = '\0'; return distro; } /** *@brief * Ensure that the shared object exists and * get the shared object name from the table * * @par Functionality: * This function checks verified and tested list of * , , and * if the above entries matches with libjob_support table, * then returns shared object to the caller for dlopen. * Otherwise it returns NULL. * * @see * ck_acct_facility_present * * @return char * * @retval libjob_support[idx].libjobver * @retval NULL: Failed to get the supported library * * @par Side Effects: None * * @par MT-safe: Yes * */ char * get_versioned_libname() { int idx; int table_size; struct utsname buf; struct libjob_support jobobj; memset(&jobobj, 0, sizeof(jobobj)); /* find OS information - loop to find out which file available */ table_size = sizeof(release_info) / sizeof(release_info[0]); for (idx = 1; idx < table_size; idx++) { if (access(release_info[idx].file, R_OK) != -1) break; } /* if we found a readable os release file, parse it. * if we dont find a file or if parse_sysfile_info fails, * jobobj.osver remains NULL, and is handled later */ if (idx < table_size) jobobj.osver = parse_sysfile_info(release_info[idx].file, release_info[idx].pfx, release_info[idx].srch, release_info[idx].sep); /* Get the information on architecture */ if (uname(&buf) == -1) { sprintf(log_buffer, "uname() call failed"); log_err(errno, __func__, log_buffer); goto SYSFAIL; } jobobj.arch = strdup(buf.machine); /* check that all the required members of jobobj are NON-NULL */ if ((jobobj.arch == NULL) || (jobobj.osver == NULL)) { sprintf(log_buffer, "Failed to get system information"); log_err(errno, __func__, log_buffer); goto SYSFAIL; } /* Compare system information with verified list of platforms */ table_size = sizeof(libjob_support) / sizeof(libjob_support[0]); for (idx = 0; idx < table_size; idx++) { if ((strcmp(jobobj.osver, libjob_support[idx].osver) == 0) && (strcmp(jobobj.arch, libjob_support[idx].arch) == 0)) { free(jobobj.arch); free(jobobj.osver); return libjob_support[idx].libjobver; } } SYSFAIL: if (jobobj.arch) free(jobobj.arch); if (jobobj.osver) free(jobobj.osver); return NULL; }