Annotation of linux/kernel/exit.c, revision 1.1.1.10

1.1.1.2   root        1: /*
                      2:  *  linux/kernel/exit.c
                      3:  *
                      4:  *  (C) 1991  Linus Torvalds
                      5:  */
                      6: 
1.1.1.4   root        7: #define DEBUG_PROC_TREE
                      8: 
1.1       root        9: #include <errno.h>
                     10: #include <signal.h>
                     11: #include <sys/wait.h>
                     12: 
                     13: #include <linux/sched.h>
                     14: #include <linux/kernel.h>
1.1.1.10! root       15: #include <linux/mm.h>
1.1       root       16: #include <linux/tty.h>
                     17: #include <asm/segment.h>
                     18: 
                     19: int sys_close(int fd);
                     20: 
1.1.1.8   root       21: int send_sig(long sig,struct task_struct * p,int priv)
1.1.1.5   root       22: {
                     23:        if (!p || (sig < 0) || (sig > 32))
                     24:                return -EINVAL;
1.1.1.8   root       25:        if (!priv && ((sig != SIGCONT) || (current->session != p->session)) &&
1.1.1.9   root       26:            (current->euid != p->euid) && (current->uid != p->uid) && !suser())
1.1.1.5   root       27:                return -EPERM;
                     28:        if (!sig)
                     29:                return 0;
                     30:        if ((sig == SIGKILL) || (sig == SIGCONT)) {
                     31:                if (p->state == TASK_STOPPED)
                     32:                        p->state = TASK_RUNNING;
                     33:                p->exit_code = 0;
                     34:                p->signal &= ~( (1<<(SIGSTOP-1)) | (1<<(SIGTSTP-1)) |
                     35:                                (1<<(SIGTTIN-1)) | (1<<(SIGTTOU-1)) );
                     36:        } 
                     37:        /* Depends on order SIGSTOP, SIGTSTP, SIGTTIN, SIGTTOU */
                     38:        if ((sig >= SIGSTOP) && (sig <= SIGTTOU)) 
                     39:                p->signal &= ~(1<<(SIGCONT-1));
                     40:        /* Actually deliver the signal */
                     41:        p->signal |= (1<<(sig-1));
                     42:        if (p->flags & PF_PTRACED) {
                     43:                /* save the signal number for wait. */
                     44:                p->exit_code = sig;
                     45: 
1.1.1.9   root       46:                /* we have to make sure the parent process is awake. */
1.1.1.5   root       47:                if (p->p_pptr != NULL && p->p_pptr->state == TASK_INTERRUPTIBLE)
                     48:                        p->p_pptr->state = TASK_RUNNING;
                     49: 
                     50:                /* we have to make sure that the process stops. */
                     51:                if (p->state == TASK_INTERRUPTIBLE || p->state == TASK_RUNNING)
                     52:                        p->state = TASK_STOPPED;
                     53:        }
                     54:        return 0;
                     55: }
                     56: 
1.1       root       57: void release(struct task_struct * p)
                     58: {
                     59:        int i;
                     60: 
                     61:        if (!p)
                     62:                return;
1.1.1.4   root       63:        if (p == current) {
                     64:                printk("task releasing itself\n\r");
                     65:                return;
                     66:        }
1.1       root       67:        for (i=1 ; i<NR_TASKS ; i++)
1.1.1.7   root       68:                if (task[i] == p) {
                     69:                        task[i] = NULL;
1.1.1.9   root       70:                        REMOVE_LINKS(p);
1.1.1.7   root       71:                        free_page((long) p);
1.1       root       72:                        return;
                     73:                }
                     74:        panic("trying to release non-existent task");
                     75: }
                     76: 
1.1.1.4   root       77: #ifdef DEBUG_PROC_TREE
                     78: /*
                     79:  * Check to see if a task_struct pointer is present in the task[] array
                     80:  * Return 0 if found, and 1 if not found.
                     81:  */
                     82: int bad_task_ptr(struct task_struct *p)
                     83: {
                     84:        int     i;
                     85: 
                     86:        if (!p)
                     87:                return 0;
                     88:        for (i=0 ; i<NR_TASKS ; i++)
                     89:                if (task[i] == p)
                     90:                        return 0;
                     91:        return 1;
                     92: }
                     93:        
                     94: /*
                     95:  * This routine scans the pid tree and make sure the rep invarient still
                     96:  * holds.  Used for debugging only, since it's very slow....
                     97:  *
                     98:  * It looks a lot scarier than it really is.... we're doing ænothing more
                     99:  * than verifying the doubly-linked list foundæin p_ysptr and p_osptr, 
                    100:  * and checking it corresponds with the process tree defined by p_cptr and 
                    101:  * p_pptr;
                    102:  */
                    103: void audit_ptree()
                    104: {
                    105:        int     i;
                    106: 
                    107:        for (i=1 ; i<NR_TASKS ; i++) {
                    108:                if (!task[i])
                    109:                        continue;
                    110:                if (bad_task_ptr(task[i]->p_pptr))
                    111:                        printk("Warning, pid %d's parent link is bad\n",
                    112:                                task[i]->pid);
                    113:                if (bad_task_ptr(task[i]->p_cptr))
                    114:                        printk("Warning, pid %d's child link is bad\n",
                    115:                                task[i]->pid);
                    116:                if (bad_task_ptr(task[i]->p_ysptr))
                    117:                        printk("Warning, pid %d's ys link is bad\n",
                    118:                                task[i]->pid);
                    119:                if (bad_task_ptr(task[i]->p_osptr))
                    120:                        printk("Warning, pid %d's os link is bad\n",
                    121:                                task[i]->pid);
                    122:                if (task[i]->p_pptr == task[i])
                    123:                        printk("Warning, pid %d parent link points to self\n");
                    124:                if (task[i]->p_cptr == task[i])
                    125:                        printk("Warning, pid %d child link points to self\n");
                    126:                if (task[i]->p_ysptr == task[i])
                    127:                        printk("Warning, pid %d ys link points to self\n");
                    128:                if (task[i]->p_osptr == task[i])
                    129:                        printk("Warning, pid %d os link points to self\n");
                    130:                if (task[i]->p_osptr) {
                    131:                        if (task[i]->p_pptr != task[i]->p_osptr->p_pptr)
                    132:                                printk(
                    133:                        "Warning, pid %d older sibling %d parent is %d\n",
                    134:                                task[i]->pid, task[i]->p_osptr->pid,
                    135:                                task[i]->p_osptr->p_pptr->pid);
                    136:                        if (task[i]->p_osptr->p_ysptr != task[i])
                    137:                                printk(
                    138:                "Warning, pid %d older sibling %d has mismatched ys link\n",
                    139:                                task[i]->pid, task[i]->p_osptr->pid);
                    140:                }
                    141:                if (task[i]->p_ysptr) {
                    142:                        if (task[i]->p_pptr != task[i]->p_ysptr->p_pptr)
                    143:                                printk(
                    144:                        "Warning, pid %d younger sibling %d parent is %d\n",
                    145:                                task[i]->pid, task[i]->p_osptr->pid,
                    146:                                task[i]->p_osptr->p_pptr->pid);
                    147:                        if (task[i]->p_ysptr->p_osptr != task[i])
                    148:                                printk(
                    149:                "Warning, pid %d younger sibling %d has mismatched os link\n",
                    150:                                task[i]->pid, task[i]->p_ysptr->pid);
                    151:                }
                    152:                if (task[i]->p_cptr) {
                    153:                        if (task[i]->p_cptr->p_pptr != task[i])
                    154:                                printk(
                    155:                        "Warning, pid %d youngest child %d has mismatched parent link\n",
                    156:                                task[i]->pid, task[i]->p_cptr->pid);
                    157:                        if (task[i]->p_cptr->p_ysptr)
                    158:                                printk(
                    159:                        "Warning, pid %d youngest child %d has non-NULL ys link\n",
                    160:                                task[i]->pid, task[i]->p_cptr->pid);
                    161:                }
                    162:        }
                    163: }
                    164: #endif /* DEBUG_PROC_TREE */
                    165: 
1.1.1.8   root      166: /*
                    167:  * This checks not only the pgrp, but falls back on the pid if no
                    168:  * satisfactory prgp is found. I dunno - gdb doesn't work correctly
                    169:  * without this...
                    170:  */
1.1.1.4   root      171: int session_of_pgrp(int pgrp)
1.1.1.2   root      172: {
1.1.1.4   root      173:        struct task_struct **p;
1.1.1.8   root      174:        int fallback;
1.1.1.4   root      175: 
1.1.1.8   root      176:        fallback = -1;
                    177:        for (p = &LAST_TASK ; p > &FIRST_TASK ; --p) {
                    178:                if (!*p || (*p)->session <= 0)
                    179:                        continue;
1.1.1.4   root      180:                if ((*p)->pgrp == pgrp)
1.1.1.8   root      181:                        return (*p)->session;
                    182:                if ((*p)->pid == pgrp)
                    183:                        fallback = (*p)->session;
                    184:        }
                    185:        return fallback;
1.1.1.4   root      186: }
                    187: 
                    188: int kill_pg(int pgrp, int sig, int priv)
                    189: {
                    190:        struct task_struct **p;
                    191:        int err,retval = -ESRCH;
                    192:        int found = 0;
                    193: 
1.1.1.5   root      194:        if (sig<0 || sig>32 || pgrp<=0)
1.1.1.4   root      195:                return -EINVAL;
                    196:        for (p = &LAST_TASK ; p > &FIRST_TASK ; --p)
1.1.1.8   root      197:                if (*p && (*p)->pgrp == pgrp) {
1.1.1.4   root      198:                        if (sig && (err = send_sig(sig,*p,priv)))
                    199:                                retval = err;
                    200:                        else
                    201:                                found++;
                    202:                }
                    203:        return(found ? 0 : retval);
                    204: }
                    205: 
                    206: int kill_proc(int pid, int sig, int priv)
                    207: {
                    208:        struct task_struct **p;
                    209: 
1.1.1.5   root      210:        if (sig<0 || sig>32)
1.1.1.4   root      211:                return -EINVAL;
                    212:        for (p = &LAST_TASK ; p > &FIRST_TASK ; --p)
1.1.1.8   root      213:                if (*p && (*p)->pid == pid)
1.1.1.4   root      214:                        return(sig ? send_sig(sig,*p,priv) : 0);
                    215:        return(-ESRCH);
1.1.1.2   root      216: }
                    217: 
                    218: /*
1.1.1.4   root      219:  * POSIX specifies that kill(-1,sig) is unspecified, but what we have
                    220:  * is probably wrong.  Should make it like BSD or SYSV.
1.1.1.2   root      221:  */
                    222: int sys_kill(int pid,int sig)
1.1       root      223: {
                    224:        struct task_struct **p = NR_TASKS + task;
1.1.1.7   root      225:        int err, retval = 0, count = 0;
1.1       root      226: 
1.1.1.4   root      227:        if (!pid)
1.1.1.7   root      228:                return(kill_pg(current->pgrp,sig,0));
1.1.1.4   root      229:        if (pid == -1) {
                    230:                while (--p > &FIRST_TASK)
1.1.1.8   root      231:                        if (*p && (*p)->pid > 1 && *p != current) {
1.1.1.7   root      232:                                ++count;
                    233:                                if ((err = send_sig(sig,*p,0)) != -EPERM)
                    234:                                        retval = err;
                    235:                        }
                    236:                return(count ? retval : -ESRCH);
1.1.1.4   root      237:        }
                    238:        if (pid < 0) 
                    239:                return(kill_pg(-pid,sig,0));
                    240:        /* Normal kill */
                    241:        return(kill_proc(pid,sig,0));
1.1       root      242: }
                    243: 
1.1.1.4   root      244: /*
                    245:  * Determine if a process group is "orphaned", according to the POSIX
                    246:  * definition in 2.2.2.52.  Orphaned process groups are not to be affected
                    247:  * by terminal-generated stop signals.  Newly orphaned process groups are 
                    248:  * to receive a SIGHUP and a SIGCONT.
                    249:  * 
                    250:  * "I ask you, have you ever known what it is to be an orphan?"
                    251:  */
                    252: int is_orphaned_pgrp(int pgrp)
1.1       root      253: {
1.1.1.4   root      254:        struct task_struct **p;
1.1.1.2   root      255: 
1.1.1.4   root      256:        for (p = &LAST_TASK ; p > &FIRST_TASK ; --p) {
                    257:                if (!(*p) ||
                    258:                    ((*p)->pgrp != pgrp) || 
                    259:                    ((*p)->state == TASK_ZOMBIE) ||
                    260:                    ((*p)->p_pptr->pid == 1))
                    261:                        continue;
                    262:                if (((*p)->p_pptr->pgrp != pgrp) &&
                    263:                    ((*p)->p_pptr->session == (*p)->session))
                    264:                        return 0;
                    265:        }
                    266:        return(1);      /* (sighing) "Often!" */
1.1       root      267: }
                    268: 
1.1.1.4   root      269: static int has_stopped_jobs(int pgrp)
1.1       root      270: {
1.1.1.4   root      271:        struct task_struct ** p;
                    272: 
                    273:        for (p = &LAST_TASK ; p > &FIRST_TASK ; --p) {
1.1.1.8   root      274:                if (!*p || (*p)->pgrp != pgrp)
1.1.1.4   root      275:                        continue;
                    276:                if ((*p)->state == TASK_STOPPED)
                    277:                        return(1);
                    278:        }
                    279:        return(0);
                    280: }
                    281: 
1.1.1.9   root      282: static void forget_original_parent(struct task_struct * father)
                    283: {
                    284:        struct task_struct ** p;
                    285: 
                    286:        for (p = &LAST_TASK ; p > &FIRST_TASK ; --p)
                    287:                if (*p && (*p)->p_opptr == father)
1.1.1.10! root      288:                        if (task[1])
        !           289:                                (*p)->p_opptr = task[1];
        !           290:                        else
        !           291:                                (*p)->p_opptr = task[0];
1.1.1.9   root      292: }
                    293: 
1.1.1.4   root      294: volatile void do_exit(long code)
                    295: {
                    296:        struct task_struct *p;
1.1       root      297:        int i;
                    298: 
1.1.1.10! root      299: fake_volatile:
1.1       root      300:        free_page_tables(get_base(current->ldt[1]),get_limit(0x0f));
                    301:        free_page_tables(get_base(current->ldt[2]),get_limit(0x17));
                    302:        for (i=0 ; i<NR_OPEN ; i++)
                    303:                if (current->filp[i])
                    304:                        sys_close(i);
1.1.1.9   root      305:        forget_original_parent(current);
1.1       root      306:        iput(current->pwd);
1.1.1.4   root      307:        current->pwd = NULL;
1.1       root      308:        iput(current->root);
1.1.1.4   root      309:        current->root = NULL;
1.1.1.3   root      310:        iput(current->executable);
1.1.1.4   root      311:        current->executable = NULL;
1.1.1.8   root      312:        for (i=0; i < current->numlibraries; i++) {
                    313:                iput(current->libraries[i].library);
                    314:                current->libraries[i].library = NULL;
                    315:        }       
1.1.1.2   root      316:        current->state = TASK_ZOMBIE;
                    317:        current->exit_code = code;
1.1.1.7   root      318:        current->rss = 0;
1.1.1.4   root      319:        /* 
                    320:         * Check to see if any process groups have become orphaned
                    321:         * as a result of our exiting, and if they have any stopped
                    322:         * jobs, send them a SIGUP and then a SIGCONT.  (POSIX 3.2.2.2)
                    323:         *
                    324:         * Case i: Our father is in a different pgrp than we are
                    325:         * and we were the only connection outside, so our pgrp
                    326:         * is about to become orphaned.
                    327:         */
                    328:        if ((current->p_pptr->pgrp != current->pgrp) &&
                    329:            (current->p_pptr->session == current->session) &&
                    330:            is_orphaned_pgrp(current->pgrp) &&
                    331:            has_stopped_jobs(current->pgrp)) {
                    332:                kill_pg(current->pgrp,SIGHUP,1);
                    333:                kill_pg(current->pgrp,SIGCONT,1);
                    334:        }
                    335:        /* Let father know we died */
1.1.1.5   root      336:        send_sig (SIGCHLD, current->p_pptr, 1);
1.1.1.4   root      337:        
                    338:        /*
                    339:         * This loop does two things:
                    340:         * 
                    341:         * A.  Make init inherit all the child processes
                    342:         * B.  Check to see if any process groups have become orphaned
                    343:         *      as a result of our exiting, and if they have any stopped
1.1.1.9   root      344:         *      jobs, send them a SIGHUP and then a SIGCONT.  (POSIX 3.2.2.2)
1.1.1.4   root      345:         */
1.1.1.7   root      346:        while (p = current->p_cptr) {
                    347:                current->p_cptr = p->p_osptr;
                    348:                p->p_ysptr = NULL;
1.1.1.9   root      349:                p->flags &= ~PF_PTRACED;
1.1.1.10! root      350:                if (task[1])
        !           351:                        p->p_pptr = task[1];
        !           352:                else
        !           353:                        p->p_pptr = task[0];
1.1.1.9   root      354:                p->p_osptr = p->p_pptr->p_cptr;
                    355:                p->p_osptr->p_ysptr = p;
                    356:                p->p_pptr->p_cptr = p;
1.1.1.7   root      357:                if (p->state == TASK_ZOMBIE)
1.1.1.10! root      358:                        send_sig(SIGCHLD,p->p_pptr,1);
1.1.1.7   root      359:                /*
                    360:                 * process group orphan check
                    361:                 * Case ii: Our child is in a different pgrp 
                    362:                 * than we are, and it was the only connection
                    363:                 * outside, so the child pgrp is now orphaned.
                    364:                 */
                    365:                if ((p->pgrp != current->pgrp) &&
                    366:                    (p->session == current->session) &&
                    367:                    is_orphaned_pgrp(p->pgrp) &&
                    368:                    has_stopped_jobs(p->pgrp)) {
                    369:                        kill_pg(p->pgrp,SIGHUP,1);
                    370:                        kill_pg(p->pgrp,SIGCONT,1);
1.1.1.4   root      371:                }
                    372:        }
                    373:        if (current->leader) {
                    374:                struct task_struct **p;
                    375:                struct tty_struct *tty;
                    376: 
                    377:                if (current->tty >= 0) {
                    378:                        tty = TTY_TABLE(current->tty);
1.1.1.6   root      379:                        if (tty->pgrp > 0)
1.1.1.4   root      380:                                kill_pg(tty->pgrp, SIGHUP, 1);
1.1.1.6   root      381:                        tty->pgrp = -1;
1.1.1.4   root      382:                        tty->session = 0;
                    383:                }
                    384:                for (p = &LAST_TASK ; p > &FIRST_TASK ; --p)
1.1.1.8   root      385:                        if (*p && (*p)->session == current->session)
1.1.1.4   root      386:                                (*p)->tty = -1;
                    387:        }
                    388:        if (last_task_used_math == current)
                    389:                last_task_used_math = NULL;
                    390: #ifdef DEBUG_PROC_TREE
                    391:        audit_ptree();
                    392: #endif
1.1       root      393:        schedule();
1.1.1.10! root      394: /*
        !           395:  * In order to get rid of the "volatile function does return" message
        !           396:  * I did this little loop that confuses gcc to think do_exit really
        !           397:  * is volatile. In fact it's schedule() that is volatile in some
        !           398:  * circumstances: when current->state = ZOMBIE, schedule() never
        !           399:  * returns.
        !           400:  *
        !           401:  * In fact the natural way to do all this is to have the label and the
        !           402:  * goto right after each other, but I put the fake_volatile label at
        !           403:  * the start of the function just in case something /really/ bad
        !           404:  * happens, and the schedule returns. This way we can try again. I'm
        !           405:  * not paranoid: it's just that everybody is out to get me.
        !           406:  */
        !           407:        goto fake_volatile;
1.1       root      408: }
                    409: 
                    410: int sys_exit(int error_code)
                    411: {
1.1.1.4   root      412:        do_exit((error_code&0xff)<<8);
1.1       root      413: }
                    414: 
1.1.1.2   root      415: int sys_waitpid(pid_t pid,unsigned long * stat_addr, int options)
1.1       root      416: {
1.1.1.4   root      417:        int flag;
                    418:        struct task_struct *p;
                    419:        unsigned long oldblocked;
1.1       root      420: 
1.1.1.5   root      421:        if (stat_addr)
                    422:                verify_area(stat_addr,4);
1.1       root      423: repeat:
1.1.1.2   root      424:        flag=0;
1.1.1.9   root      425:        for (p = current->p_cptr ; p ; p = p->p_osptr) {
1.1.1.2   root      426:                if (pid>0) {
1.1.1.4   root      427:                        if (p->pid != pid)
1.1.1.2   root      428:                                continue;
                    429:                } else if (!pid) {
1.1.1.4   root      430:                        if (p->pgrp != current->pgrp)
1.1.1.2   root      431:                                continue;
                    432:                } else if (pid != -1) {
1.1.1.4   root      433:                        if (p->pgrp != -pid)
1.1.1.2   root      434:                                continue;
                    435:                }
1.1.1.4   root      436:                switch (p->state) {
1.1.1.2   root      437:                        case TASK_STOPPED:
1.1.1.8   root      438:                                if (!p->exit_code)
                    439:                                        continue;
                    440:                                if (!(options & WUNTRACED) && !(p->flags & PF_PTRACED))
1.1.1.2   root      441:                                        continue;
1.1.1.5   root      442:                                if (stat_addr)
                    443:                                        put_fs_long((p->exit_code << 8) | 0x7f,
                    444:                                                stat_addr);
1.1.1.4   root      445:                                p->exit_code = 0;
                    446:                                return p->pid;
1.1.1.2   root      447:                        case TASK_ZOMBIE:
1.1.1.7   root      448:                                current->cutime += p->utime + p->cutime;
                    449:                                current->cstime += p->stime + p->cstime;
                    450:                                current->cmin_flt += p->min_flt + p->cmin_flt;
                    451:                                current->cmaj_flt += p->maj_flt + p->cmaj_flt;
1.1.1.4   root      452:                                flag = p->pid;
1.1.1.5   root      453:                                if (stat_addr)
                    454:                                        put_fs_long(p->exit_code, stat_addr);
1.1.1.9   root      455:                                if (p->p_opptr != p->p_pptr) {
                    456:                                        REMOVE_LINKS(p);
                    457:                                        p->p_pptr = p->p_opptr;
                    458:                                        SET_LINKS(p);
                    459:                                        send_sig(SIGCHLD,p->p_pptr,1);
                    460:                                } else
                    461:                                        release(p);
1.1.1.4   root      462: #ifdef DEBUG_PROC_TREE
                    463:                                audit_ptree();
                    464: #endif
1.1.1.2   root      465:                                return flag;
                    466:                        default:
1.1       root      467:                                flag=1;
1.1.1.2   root      468:                                continue;
                    469:                }
                    470:        }
1.1       root      471:        if (flag) {
                    472:                if (options & WNOHANG)
                    473:                        return 0;
1.1.1.2   root      474:                current->state=TASK_INTERRUPTIBLE;
1.1.1.4   root      475:                oldblocked = current->blocked;
                    476:                current->blocked &= ~(1<<(SIGCHLD-1));
1.1.1.2   root      477:                schedule();
1.1.1.4   root      478:                current->blocked = oldblocked;
                    479:                if (current->signal & ~(current->blocked | (1<<(SIGCHLD-1))))
                    480:                        return -ERESTARTSYS;
1.1       root      481:                else
1.1.1.4   root      482:                        goto repeat;
1.1       root      483:        }
                    484:        return -ECHILD;
                    485: }

unix.superglobalmegacorp.com