int procfs_doprocctl(PFS_FILL_ARGS) { int error; struct namemap *nm; if (uio == NULL || uio->uio_rw != UIO_WRITE) return (EOPNOTSUPP); /* * Map signal names into signal generation * or debug control. Unknown commands and/or signals * return EOPNOTSUPP. * * Sending a signal while the process is being debugged * also has the side effect of letting the target continue * to run. There is no way to single-step a signal delivery. */ error = EOPNOTSUPP; sbuf_trim(sb); sbuf_finish(sb); nm = findname(ctlnames, sbuf_data(sb), sbuf_len(sb)); if (nm) { printf("procfs: got a %s command\n", sbuf_data(sb)); error = procfs_control(td, p, nm->nm_val); } else { nm = findname(signames, sbuf_data(sb), sbuf_len(sb)); if (nm) { printf("procfs: got a sig%s\n", sbuf_data(sb)); PROC_LOCK(p); /* This is very broken XXXKSE: */ if (TRACE_WAIT_P(td->td_proc, p)) { p->p_xstat = nm->nm_val; #ifdef FIX_SSTEP /* XXXKSE: */ FIX_SSTEP(FIRST_THREAD_IN_PROC(p)); #endif /* XXXKSE: */ p->p_flag &= ~P_STOPPED_SIG; PROC_SLOCK(p); thread_unsuspend(p); PROC_SUNLOCK(p); } else psignal(p, nm->nm_val); PROC_UNLOCK(p); error = 0; } } return (error); }
int sys_sched_setparam(struct thread *td, struct sched_setparam_args *uap) { struct thread *targettd; struct proc *targetp; int e; struct sched_param sched_param; e = copyin(uap->param, &sched_param, sizeof(sched_param)); if (e) return (e); if (uap->pid == 0) { targetp = td->td_proc; targettd = td; PROC_LOCK(targetp); } else { targetp = pfind(uap->pid); if (targetp == NULL) return (ESRCH); targettd = FIRST_THREAD_IN_PROC(targetp); } e = p_cansched(td, targetp); if (e == 0) { e = ksched_setparam(ksched, targettd, (const struct sched_param *)&sched_param); } PROC_UNLOCK(targetp); return (e); }
int sys_sched_getparam(struct thread *td, struct sched_getparam_args *uap) { int e; struct sched_param sched_param; struct thread *targettd; struct proc *targetp; if (uap->pid == 0) { targetp = td->td_proc; targettd = td; PROC_LOCK(targetp); } else { targetp = pfind(uap->pid); if (targetp == NULL) { return (ESRCH); } targettd = FIRST_THREAD_IN_PROC(targetp); } e = p_cansee(td, targetp); if (e == 0) { e = ksched_getparam(ksched, targettd, &sched_param); } PROC_UNLOCK(targetp); if (e == 0) e = copyout(&sched_param, uap->param, sizeof(sched_param)); return (e); }
int kproc_kthread_add(void (*func)(void *), void *arg, struct proc **procptr, struct thread **tdptr, int flags, int pages, const char *procname, const char *fmt, ...) { int error; va_list ap; char buf[100]; struct thread *td; if (*procptr == 0) { error = kproc_create(func, arg, procptr, flags, pages, "%s", procname); if (error) return (error); td = FIRST_THREAD_IN_PROC(*procptr); if (tdptr) *tdptr = td; va_start(ap, fmt); vsnprintf(td->td_name, sizeof(td->td_name), fmt, ap); va_end(ap); #ifdef KTR sched_clear_tdname(td); #endif return (0); } va_start(ap, fmt); vsnprintf(buf, sizeof(buf), fmt, ap); va_end(ap); error = kthread_add(func, arg, *procptr, tdptr, flags, pages, "%s", buf); return (error); }
int linux_fork(struct thread *td, struct linux_fork_args *args) { struct fork_req fr; int error; struct proc *p2; struct thread *td2; #ifdef DEBUG if (ldebug(fork)) printf(ARGS(fork, "")); #endif bzero(&fr, sizeof(fr)); fr.fr_flags = RFFDG | RFPROC | RFSTOPPED; fr.fr_procp = &p2; if ((error = fork1(td, &fr)) != 0) return (error); td2 = FIRST_THREAD_IN_PROC(p2); linux_proc_init(td, td2, 0); td->td_retval[0] = p2->p_pid; /* * Make this runnable after we are finished with it. */ thread_lock(td2); TD_SET_CAN_RUN(td2); sched_add(td2, SRQ_BORING); thread_unlock(td2); return (0); }
static void ald_daemon(void) { int needwakeup; struct alq *alq; ald_thread = FIRST_THREAD_IN_PROC(ald_proc); EVENTHANDLER_REGISTER(shutdown_pre_sync, ald_shutdown, NULL, SHUTDOWN_PRI_FIRST); ALD_LOCK(); for (;;) { while ((alq = LIST_FIRST(&ald_active)) == NULL) msleep(&ald_active, &ald_mtx, PWAIT, "aldslp", 0); ALQ_LOCK(alq); ald_deactivate(alq); ALD_UNLOCK(); needwakeup = alq_doio(alq); ALQ_UNLOCK(alq); if (needwakeup) wakeup(alq); ALD_LOCK(); } }
int kern_sched_rr_get_interval(struct thread *td, pid_t pid, struct timespec *ts) { int e; struct thread *targettd; struct proc *targetp; if (pid == 0) { targettd = td; targetp = td->td_proc; PROC_LOCK(targetp); } else { targetp = pfind(pid); if (targetp == NULL) return (ESRCH); targettd = FIRST_THREAD_IN_PROC(targetp); } e = p_cansee(td, targetp); if (e == 0) e = ksched_rr_get_interval(ksched, targettd, ts); PROC_UNLOCK(targetp); return (e); }
int sys_sched_getscheduler(struct thread *td, struct sched_getscheduler_args *uap) { int e, policy; struct thread *targettd; struct proc *targetp; if (uap->pid == 0) { targetp = td->td_proc; targettd = td; PROC_LOCK(targetp); } else { targetp = pfind(uap->pid); if (targetp == NULL) return (ESRCH); targettd = FIRST_THREAD_IN_PROC(targetp); } e = p_cansee(td, targetp); if (e == 0) { e = ksched_getscheduler(ksched, targettd, &policy); td->td_retval[0] = policy; } PROC_UNLOCK(targetp); return (e); }
int linux_fork(struct thread *td, struct linux_fork_args *args) { int error; struct proc *p2; struct thread *td2; #ifdef DEBUG if (ldebug(fork)) printf(ARGS(fork, "")); #endif if ((error = fork1(td, RFFDG | RFPROC | RFSTOPPED, 0, &p2, NULL, 0)) != 0) return (error); td->td_retval[0] = p2->p_pid; td->td_retval[1] = 0; error = linux_proc_init(td, td->td_retval[0], 0); if (error) return (error); td2 = FIRST_THREAD_IN_PROC(p2); /* * Make this runnable after we are finished with it. */ thread_lock(td2); TD_SET_CAN_RUN(td2); sched_add(td2, SRQ_BORING); thread_unlock(td2); return (0); }
static int exec_linux_imgact_try(struct image_params *imgp) { const char *head = (const char *)imgp->image_header; char *rpath; int error = -1, len; /* * The interpreter for shell scripts run from a linux binary needs * to be located in /compat/linux if possible in order to recursively * maintain linux path emulation. */ if (((const short *)head)[0] == SHELLMAGIC) { /* * Run our normal shell image activator. If it succeeds * attempt to use the alternate path for the interpreter. * If an alternate path is found, use our stringspace * to store it. */ if ((error = exec_shell_imgact(imgp)) == 0) { linux_emul_convpath(FIRST_THREAD_IN_PROC(imgp->proc), imgp->interpreter_name, UIO_SYSSPACE, &rpath, 0, AT_FDCWD); if (rpath != NULL) { len = strlen(rpath) + 1; if (len <= MAXSHELLCMDLEN) memcpy(imgp->interpreter_name, rpath, len); free(rpath, M_TEMP); } } } return(error); }
static void tesla_perthread_process_dtor(__unused void *arg, struct proc *p) { struct thread *td; td = FIRST_THREAD_IN_PROC(p); if (td != NULL && td->td_tesla != NULL) tesla_store_reset(td->td_tesla); }
int procfs_doprocfpregs(PFS_FILL_ARGS) { int error; struct fpreg r; struct thread *td2; #ifdef COMPAT_FREEBSD32 struct fpreg32 r32; int wrap32 = 0; #endif if (uio->uio_offset != 0) return (0); PROC_LOCK(p); PROC_ASSERT_HELD(p); if (p_candebug(td, p)) { PROC_UNLOCK(p); return (EPERM); } if (!P_SHOULDSTOP(p)) { PROC_UNLOCK(p); return (EBUSY); } td2 = FIRST_THREAD_IN_PROC(p); #ifdef COMPAT_FREEBSD32 if (SV_CURPROC_FLAG(SV_ILP32)) { if (SV_PROC_FLAG(td2->td_proc, SV_ILP32) == 0) { PROC_UNLOCK(p); return (EINVAL); } wrap32 = 1; memset(&r32, 0, sizeof(r32)); } else #endif memset(&r, 0, sizeof(r)); error = PROC(read, fpregs, td2, &r); if (error == 0) { PROC_UNLOCK(p); error = UIOMOVE_FROMBUF(r, uio); PROC_LOCK(p); } if (error == 0 && uio->uio_rw == UIO_WRITE) { if (!P_SHOULDSTOP(p)) error = EBUSY; else /* XXXKSE: */ error = PROC(write, fpregs, td2, &r); } PROC_UNLOCK(p); return (error); }
static void g_up_procbody(void) { struct proc *p = g_up_proc; struct thread *tp = FIRST_THREAD_IN_PROC(p); mtx_assert(&Giant, MA_NOTOWNED); thread_lock(tp); sched_prio(tp, PRIBIO); thread_unlock(tp); for(;;) { g_io_schedule_up(tp); } }
static void g_event_procbody(void) { struct proc *p = g_event_proc; struct thread *tp = FIRST_THREAD_IN_PROC(p); mtx_assert(&Giant, MA_NOTOWNED); thread_lock(tp); sched_prio(tp, PRIBIO); thread_unlock(tp); for(;;) { g_run_events(); tsleep(&g_wait_event, PRIBIO, "-", hz/10); } }
int procfs_doprocdbregs(PFS_FILL_ARGS) { int error; struct dbreg r; struct thread *td2; #ifdef COMPAT_FREEBSD32 struct dbreg32 r32; int wrap32 = 0; #endif if (uio->uio_offset != 0) return (0); PROC_LOCK(p); KASSERT(p->p_lock > 0, ("proc not held")); if (p_candebug(td, p) != 0) { PROC_UNLOCK(p); return (EPERM); } td2 = FIRST_THREAD_IN_PROC(p); #ifdef COMPAT_FREEBSD32 if (SV_CURPROC_FLAG(SV_ILP32)) { if (SV_PROC_FLAG(td2->td_proc, SV_ILP32) == 0) { PROC_UNLOCK(p); return (EINVAL); } wrap32 = 1; } #endif error = PROC(read, dbregs, td2, &r); if (error == 0) { PROC_UNLOCK(p); error = UIOMOVE_FROMBUF(r, uio); PROC_LOCK(p); } if (error == 0 && uio->uio_rw == UIO_WRITE) { if (!P_SHOULDSTOP(p)) /* XXXKSE should be P_TRACED? */ error = EBUSY; else /* XXXKSE: */ error = PROC(write, dbregs, td2, &r); } PROC_UNLOCK(p); return (error); }
/* * UMA should ensure that this function is never called. * Freeing a proc structure would violate type stability. */ static void proc_fini(void *mem, int size) { #ifdef notnow struct proc *p; p = (struct proc *)mem; EVENTHANDLER_INVOKE(process_fini, p); pstats_free(p->p_stats); thread_free(FIRST_THREAD_IN_PROC(p)); mtx_destroy(&p->p_mtx); if (p->p_ksi != NULL) ksiginfo_free(p->p_ksi); #else panic("proc reclaimed"); #endif }
int linux_vfork(struct thread *td, struct linux_vfork_args *args) { int error; struct proc *p2; struct thread *td2; #ifdef DEBUG if (ldebug(vfork)) printf(ARGS(vfork, "")); #endif /* Exclude RFPPWAIT */ if ((error = fork1(td, RFFDG | RFPROC | RFMEM | RFSTOPPED, 0, &p2, NULL, 0)) != 0) return (error); td->td_retval[0] = p2->p_pid; error = linux_proc_init(td, td->td_retval[0], 0); if (error) return (error); PROC_LOCK(p2); p2->p_flag |= P_PPWAIT; PROC_UNLOCK(p2); td2 = FIRST_THREAD_IN_PROC(p2); /* * Make this runnable after we are finished with it. */ thread_lock(td2); TD_SET_CAN_RUN(td2); sched_add(td2, SRQ_BORING); thread_unlock(td2); /* wait for the children to exit, ie. emulate vfork */ PROC_LOCK(p2); while (p2->p_flag & P_PPWAIT) cv_wait(&p2->p_pwait, &p2->p_mtx); PROC_UNLOCK(p2); return (0); }
static void pagezero_start(void __unused *arg) { int error; error = kthread_create(vm_pagezero, NULL, &pagezero_proc, RFSTOPPED, 0, "pagezero"); if (error) panic("pagezero_start: error %d\n", error); /* * We're an idle task, don't count us in the load. */ PROC_LOCK(pagezero_proc); pagezero_proc->p_flag |= P_NOLOAD; PROC_UNLOCK(pagezero_proc); mtx_lock_spin(&sched_lock); setrunqueue(FIRST_THREAD_IN_PROC(pagezero_proc), SRQ_BORING); mtx_unlock_spin(&sched_lock); }
DECLHIDDEN(int) rtThreadNativeCreate(PRTTHREADINT pThreadInt, PRTNATIVETHREAD pNativeThread) { int rc; struct proc *pProc; #if __FreeBSD_version >= 800002 rc = kproc_create(rtThreadNativeMain, pThreadInt, &pProc, RFHIGHPID, 0, "%s", pThreadInt->szName); #else rc = kthread_create(rtThreadNativeMain, pThreadInt, &pProc, RFHIGHPID, 0, "%s", pThreadInt->szName); #endif if (!rc) { *pNativeThread = (RTNATIVETHREAD)FIRST_THREAD_IN_PROC(pProc); rc = VINF_SUCCESS; } else rc = RTErrConvertFromErrno(rc); return rc; }
static void pagezero_start(void __unused *arg) { int error; struct proc *p; struct thread *td; error = kproc_create(vm_pagezero, NULL, &p, RFSTOPPED, 0, "pagezero"); if (error) panic("pagezero_start: error %d\n", error); td = FIRST_THREAD_IN_PROC(p); thread_lock(td); /* We're an idle task, don't count us in the load. */ td->td_flags |= TDF_NOLOAD; sched_class(td, PRI_IDLE); sched_prio(td, PRI_MAX_IDLE); sched_add(td, SRQ_BORING); thread_unlock(td); }
static void ald_daemon(void) { int needwakeup; struct alq *alq; ald_thread = FIRST_THREAD_IN_PROC(ald_proc); alq_eventhandler_tag = EVENTHANDLER_REGISTER(shutdown_pre_sync, ald_shutdown, NULL, SHUTDOWN_PRI_FIRST); ALD_LOCK(); for (;;) { while ((alq = BSD_LIST_FIRST(&ald_active)) == NULL && !ald_shutingdown) mtx_sleep(&ald_active, &ald_mtx, PWAIT, "aldslp", 0); /* Don't shutdown until all active ALQs are flushed. */ if (ald_shutingdown && alq == NULL) { ALD_UNLOCK(); break; } ALQ_LOCK(alq); ald_deactivate(alq); ALD_UNLOCK(); needwakeup = alq_doio(alq); ALQ_UNLOCK(alq); if (needwakeup) wakeup_one(alq); ALD_LOCK(); } kproc_exit(0); }
/* * Reclaim a proc after use. */ static void proc_dtor(void *mem, int size, void *arg) { struct proc *p; struct thread *td; /* INVARIANTS checks go here */ p = (struct proc *)mem; td = FIRST_THREAD_IN_PROC(p); SDT_PROBE(proc, kernel, dtor, entry, p, size, arg, td, 0); if (td != NULL) { #ifdef INVARIANTS KASSERT((p->p_numthreads == 1), ("bad number of threads in exiting process")); KASSERT(STAILQ_EMPTY(&p->p_ktr), ("proc_dtor: non-empty p_ktr")); #endif /* Free all OSD associated to this thread. */ osd_thread_exit(td); } EVENTHANDLER_INVOKE(process_dtor, p); if (p->p_ksi != NULL) KASSERT(! KSI_ONQ(p->p_ksi), ("SIGCHLD queue")); SDT_PROBE(proc, kernel, dtor, return, p, size, arg, 0, 0); }
int sys_sched_setscheduler(struct thread *td, struct sched_setscheduler_args *uap) { int e; struct sched_param sched_param; struct thread *targettd; struct proc *targetp; /* Don't allow non root user to set a scheduler policy. */ e = priv_check(td, PRIV_SCHED_SET); if (e) return (e); e = copyin(uap->param, &sched_param, sizeof(sched_param)); if (e) return (e); if (uap->pid == 0) { targetp = td->td_proc; targettd = td; PROC_LOCK(targetp); } else { targetp = pfind(uap->pid); if (targetp == NULL) return (ESRCH); targettd = FIRST_THREAD_IN_PROC(targetp); } e = p_cansched(td, targetp); if (e == 0) { e = ksched_setscheduler(ksched, targettd, uap->policy, (const struct sched_param *)&sched_param); } PROC_UNLOCK(targetp); return (e); }
static int linux_clone_proc(struct thread *td, struct linux_clone_args *args) { struct fork_req fr; int error, ff = RFPROC | RFSTOPPED; struct proc *p2; struct thread *td2; int exit_signal; struct linux_emuldata *em; #ifdef DEBUG if (ldebug(clone)) { printf(ARGS(clone, "flags %x, stack %p, parent tid: %p, " "child tid: %p"), (unsigned)args->flags, args->stack, args->parent_tidptr, args->child_tidptr); } #endif exit_signal = args->flags & 0x000000ff; if (LINUX_SIG_VALID(exit_signal)) { exit_signal = linux_to_bsd_signal(exit_signal); } else if (exit_signal != 0) return (EINVAL); if (args->flags & LINUX_CLONE_VM) ff |= RFMEM; if (args->flags & LINUX_CLONE_SIGHAND) ff |= RFSIGSHARE; /* * XXX: In Linux, sharing of fs info (chroot/cwd/umask) * and open files is independent. In FreeBSD, its in one * structure but in reality it does not cause any problems * because both of these flags are usually set together. */ if (!(args->flags & (LINUX_CLONE_FILES | LINUX_CLONE_FS))) ff |= RFFDG; if (args->flags & LINUX_CLONE_PARENT_SETTID) if (args->parent_tidptr == NULL) return (EINVAL); if (args->flags & LINUX_CLONE_VFORK) ff |= RFPPWAIT; bzero(&fr, sizeof(fr)); fr.fr_flags = ff; fr.fr_procp = &p2; error = fork1(td, &fr); if (error) return (error); td2 = FIRST_THREAD_IN_PROC(p2); /* create the emuldata */ linux_proc_init(td, td2, args->flags); em = em_find(td2); KASSERT(em != NULL, ("clone_proc: emuldata not found.\n")); if (args->flags & LINUX_CLONE_CHILD_SETTID) em->child_set_tid = args->child_tidptr; else em->child_set_tid = NULL; if (args->flags & LINUX_CLONE_CHILD_CLEARTID) em->child_clear_tid = args->child_tidptr; else em->child_clear_tid = NULL; if (args->flags & LINUX_CLONE_PARENT_SETTID) { error = copyout(&p2->p_pid, args->parent_tidptr, sizeof(p2->p_pid)); if (error) printf(LMSG("copyout failed!")); } PROC_LOCK(p2); p2->p_sigparent = exit_signal; PROC_UNLOCK(p2); /* * In a case of stack = NULL, we are supposed to COW calling process * stack. This is what normal fork() does, so we just keep tf_rsp arg * intact. */ linux_set_upcall_kse(td2, PTROUT(args->stack)); if (args->flags & LINUX_CLONE_SETTLS) linux_set_cloned_tls(td2, args->tls); /* * If CLONE_PARENT is set, then the parent of the new process will be * the same as that of the calling process. */ if (args->flags & LINUX_CLONE_PARENT) { sx_xlock(&proctree_lock); PROC_LOCK(p2); proc_reparent(p2, td->td_proc->p_pptr); PROC_UNLOCK(p2); sx_xunlock(&proctree_lock); } #ifdef DEBUG if (ldebug(clone)) printf(LMSG("clone: successful rfork to %d, " "stack %p sig = %d"), (int)p2->p_pid, args->stack, exit_signal); #endif /* * Make this runnable after we are finished with it. */ thread_lock(td2); TD_SET_CAN_RUN(td2); sched_add(td2, SRQ_BORING); thread_unlock(td2); td->td_retval[0] = p2->p_pid; return (0); }
int linux_clone(struct thread *td, struct linux_clone_args *args) { int error, ff = RFPROC | RFSTOPPED; struct proc *p2; struct thread *td2; int exit_signal; struct linux_emuldata *em; #ifdef DEBUG if (ldebug(clone)) { printf(ARGS(clone, "flags %x, stack %p, parent tid: %p, " "child tid: %p"), (unsigned)args->flags, args->stack, args->parent_tidptr, args->child_tidptr); } #endif exit_signal = args->flags & 0x000000ff; if (LINUX_SIG_VALID(exit_signal)) { if (exit_signal <= LINUX_SIGTBLSZ) exit_signal = linux_to_bsd_signal[_SIG_IDX(exit_signal)]; } else if (exit_signal != 0) return (EINVAL); if (args->flags & LINUX_CLONE_VM) ff |= RFMEM; if (args->flags & LINUX_CLONE_SIGHAND) ff |= RFSIGSHARE; /* * XXX: In Linux, sharing of fs info (chroot/cwd/umask) * and open files is independant. In FreeBSD, its in one * structure but in reality it does not cause any problems * because both of these flags are usually set together. */ if (!(args->flags & (LINUX_CLONE_FILES | LINUX_CLONE_FS))) ff |= RFFDG; /* * Attempt to detect when linux_clone(2) is used for creating * kernel threads. Unfortunately despite the existence of the * CLONE_THREAD flag, version of linuxthreads package used in * most popular distros as of beginning of 2005 doesn't make * any use of it. Therefore, this detection relies on * empirical observation that linuxthreads sets certain * combination of flags, so that we can make more or less * precise detection and notify the FreeBSD kernel that several * processes are in fact part of the same threading group, so * that special treatment is necessary for signal delivery * between those processes and fd locking. */ if ((args->flags & 0xffffff00) == LINUX_THREADING_FLAGS) ff |= RFTHREAD; if (args->flags & LINUX_CLONE_PARENT_SETTID) if (args->parent_tidptr == NULL) return (EINVAL); error = fork1(td, ff, 0, &p2, NULL, 0); if (error) return (error); if (args->flags & (LINUX_CLONE_PARENT | LINUX_CLONE_THREAD)) { sx_xlock(&proctree_lock); PROC_LOCK(p2); proc_reparent(p2, td->td_proc->p_pptr); PROC_UNLOCK(p2); sx_xunlock(&proctree_lock); } /* create the emuldata */ error = linux_proc_init(td, p2->p_pid, args->flags); /* reference it - no need to check this */ em = em_find(p2, EMUL_DOLOCK); KASSERT(em != NULL, ("clone: emuldata not found.")); /* and adjust it */ if (args->flags & LINUX_CLONE_THREAD) { #ifdef notyet PROC_LOCK(p2); p2->p_pgrp = td->td_proc->p_pgrp; PROC_UNLOCK(p2); #endif exit_signal = 0; } if (args->flags & LINUX_CLONE_CHILD_SETTID) em->child_set_tid = args->child_tidptr; else em->child_set_tid = NULL; if (args->flags & LINUX_CLONE_CHILD_CLEARTID) em->child_clear_tid = args->child_tidptr; else em->child_clear_tid = NULL; EMUL_UNLOCK(&emul_lock); if (args->flags & LINUX_CLONE_PARENT_SETTID) { error = copyout(&p2->p_pid, args->parent_tidptr, sizeof(p2->p_pid)); if (error) printf(LMSG("copyout failed!")); } PROC_LOCK(p2); p2->p_sigparent = exit_signal; PROC_UNLOCK(p2); td2 = FIRST_THREAD_IN_PROC(p2); /* * In a case of stack = NULL, we are supposed to COW calling process * stack. This is what normal fork() does, so we just keep tf_rsp arg * intact. */ if (args->stack) linux_set_upcall_kse(td2, PTROUT(args->stack)); if (args->flags & LINUX_CLONE_SETTLS) linux_set_cloned_tls(td2, args->tls); #ifdef DEBUG if (ldebug(clone)) printf(LMSG("clone: successful rfork to %d, " "stack %p sig = %d"), (int)p2->p_pid, args->stack, exit_signal); #endif if (args->flags & LINUX_CLONE_VFORK) { PROC_LOCK(p2); p2->p_flag |= P_PPWAIT; PROC_UNLOCK(p2); } /* * Make this runnable after we are finished with it. */ thread_lock(td2); TD_SET_CAN_RUN(td2); sched_add(td2, SRQ_BORING); thread_unlock(td2); td->td_retval[0] = p2->p_pid; td->td_retval[1] = 0; if (args->flags & LINUX_CLONE_VFORK) { /* wait for the children to exit, ie. emulate vfork */ PROC_LOCK(p2); while (p2->p_flag & P_PPWAIT) cv_wait(&p2->p_pwait, &p2->p_mtx); PROC_UNLOCK(p2); } return (0); }
int procfs_doprocstatus(PFS_FILL_ARGS) { struct session *sess; struct thread *tdfirst; struct tty *tp; struct ucred *cr; const char *wmesg; char *pc; char *sep; int pid, ppid, pgid, sid; int i; pid = p->p_pid; PROC_LOCK(p); ppid = p->p_pptr ? p->p_pptr->p_pid : 0; pgid = p->p_pgrp->pg_id; sess = p->p_pgrp->pg_session; SESS_LOCK(sess); sid = sess->s_leader ? sess->s_leader->p_pid : 0; /* comm pid ppid pgid sid tty ctty,sldr start ut st wmsg euid ruid rgid,egid,groups[1 .. ngroups] */ pc = p->p_comm; do { if (*pc < 33 || *pc > 126 || *pc == '\\') sbuf_printf(sb, "\\%03o", *pc); else sbuf_putc(sb, *pc); } while (*++pc); sbuf_printf(sb, " %d %d %d %d ", pid, ppid, pgid, sid); if ((p->p_flag & P_CONTROLT) && (tp = sess->s_ttyp)) sbuf_printf(sb, "%s ", devtoname(tp->t_dev)); else sbuf_printf(sb, "- "); sep = ""; if (sess->s_ttyvp) { sbuf_printf(sb, "%sctty", sep); sep = ","; } if (SESS_LEADER(p)) { sbuf_printf(sb, "%ssldr", sep); sep = ","; } SESS_UNLOCK(sess); if (*sep != ',') { sbuf_printf(sb, "noflags"); } tdfirst = FIRST_THREAD_IN_PROC(p); thread_lock(tdfirst); if (tdfirst->td_wchan != NULL) { KASSERT(tdfirst->td_wmesg != NULL, ("wchan %p has no wmesg", tdfirst->td_wchan)); wmesg = tdfirst->td_wmesg; } else wmesg = "nochan"; thread_unlock(tdfirst); if (p->p_flag & P_INMEM) { struct timeval start, ut, st; PROC_SLOCK(p); calcru(p, &ut, &st); PROC_SUNLOCK(p); start = p->p_stats->p_start; timevaladd(&start, &boottime); sbuf_printf(sb, " %jd,%ld %jd,%ld %jd,%ld", (intmax_t)start.tv_sec, start.tv_usec, (intmax_t)ut.tv_sec, ut.tv_usec, (intmax_t)st.tv_sec, st.tv_usec); } else sbuf_printf(sb, " -1,-1 -1,-1 -1,-1"); sbuf_printf(sb, " %s", wmesg); cr = p->p_ucred; sbuf_printf(sb, " %lu %lu %lu", (u_long)cr->cr_uid, (u_long)cr->cr_ruid, (u_long)cr->cr_rgid); /* egid (cr->cr_svgid) is equal to cr_ngroups[0] see also getegid(2) in /sys/kern/kern_prot.c */ for (i = 0; i < cr->cr_ngroups; i++) { sbuf_printf(sb, ",%lu", (u_long)cr->cr_groups[i]); } if (jailed(cr)) { mtx_lock(&cr->cr_prison->pr_mtx); sbuf_printf(sb, " %s", prison_name(td->td_ucred->cr_prison, cr->cr_prison)); mtx_unlock(&cr->cr_prison->pr_mtx); } else { sbuf_printf(sb, " -"); } PROC_UNLOCK(p); sbuf_printf(sb, "\n"); return (0); }
KASSERT(p != NULL, ("process not found in proc_init\n")); p->p_emuldata = em; PROC_UNLOCK(p); } else EMUL_UNLOCK(&emul_lock); LIN_SDT_PROBE0(emul, proc_init, return); return (0); } void linux_proc_exit(void *arg __unused, struct proc *p) { struct linux_emuldata *em; int error, shared_flags, shared_xstat; struct thread *td = FIRST_THREAD_IN_PROC(p); int *child_clear_tid; struct proc *q, *nq; if (__predict_true(p->p_sysent != &elf_linux_sysvec)) return; LIN_SDT_PROBE1(emul, proc_exit, entry, p); release_futexes(p); /* find the emuldata */ em = em_find(p, EMUL_DOLOCK); KASSERT(em != NULL, ("proc_exit: emuldata not found.\n"));
int linux_ptrace(struct thread *td, struct linux_ptrace_args *uap) { union { struct linux_pt_reg reg; struct linux_pt_fpreg fpreg; struct linux_pt_fpxreg fpxreg; } r; union { struct reg bsd_reg; struct fpreg bsd_fpreg; struct dbreg bsd_dbreg; } u; void *addr; pid_t pid; int error, req; error = 0; /* by default, just copy data intact */ req = uap->req; pid = (pid_t)uap->pid; addr = (void *)uap->addr; switch (req) { case PTRACE_TRACEME: case PTRACE_POKETEXT: case PTRACE_POKEDATA: case PTRACE_KILL: error = kern_ptrace(td, req, pid, addr, uap->data); break; case PTRACE_PEEKTEXT: case PTRACE_PEEKDATA: { /* need to preserve return value */ int rval = td->td_retval[0]; error = kern_ptrace(td, req, pid, addr, 0); if (error == 0) error = copyout(td->td_retval, (void *)uap->data, sizeof(l_int)); td->td_retval[0] = rval; break; } case PTRACE_DETACH: error = kern_ptrace(td, PT_DETACH, pid, (void *)1, map_signum(uap->data)); break; case PTRACE_SINGLESTEP: case PTRACE_CONT: error = kern_ptrace(td, req, pid, (void *)1, map_signum(uap->data)); break; case PTRACE_ATTACH: error = kern_ptrace(td, PT_ATTACH, pid, addr, uap->data); break; case PTRACE_GETREGS: /* Linux is using data where FreeBSD is using addr */ error = kern_ptrace(td, PT_GETREGS, pid, &u.bsd_reg, 0); if (error == 0) { map_regs_to_linux(&u.bsd_reg, &r.reg); error = copyout(&r.reg, (void *)uap->data, sizeof(r.reg)); } break; case PTRACE_SETREGS: /* Linux is using data where FreeBSD is using addr */ error = copyin((void *)uap->data, &r.reg, sizeof(r.reg)); if (error == 0) { map_regs_from_linux(&u.bsd_reg, &r.reg); error = kern_ptrace(td, PT_SETREGS, pid, &u.bsd_reg, 0); } break; case PTRACE_GETFPREGS: /* Linux is using data where FreeBSD is using addr */ error = kern_ptrace(td, PT_GETFPREGS, pid, &u.bsd_fpreg, 0); if (error == 0) { map_fpregs_to_linux(&u.bsd_fpreg, &r.fpreg); error = copyout(&r.fpreg, (void *)uap->data, sizeof(r.fpreg)); } break; case PTRACE_SETFPREGS: /* Linux is using data where FreeBSD is using addr */ error = copyin((void *)uap->data, &r.fpreg, sizeof(r.fpreg)); if (error == 0) { map_fpregs_from_linux(&u.bsd_fpreg, &r.fpreg); error = kern_ptrace(td, PT_SETFPREGS, pid, &u.bsd_fpreg, 0); } break; case PTRACE_SETFPXREGS: #ifdef CPU_ENABLE_SSE error = copyin((void *)uap->data, &r.fpxreg, sizeof(r.fpxreg)); if (error) break; #endif /* FALL THROUGH */ case PTRACE_GETFPXREGS: { #ifdef CPU_ENABLE_SSE struct proc *p; struct thread *td2; if (sizeof(struct linux_pt_fpxreg) != sizeof(struct savexmm)) { static int once = 0; if (!once) { printf("linux: savexmm != linux_pt_fpxreg\n"); once = 1; } error = EIO; break; } if ((p = pfind(uap->pid)) == NULL) { error = ESRCH; break; } /* Exiting processes can't be debugged. */ if ((p->p_flag & P_WEXIT) != 0) { error = ESRCH; goto fail; } if ((error = p_candebug(td, p)) != 0) goto fail; /* System processes can't be debugged. */ if ((p->p_flag & P_SYSTEM) != 0) { error = EINVAL; goto fail; } /* not being traced... */ if ((p->p_flag & P_TRACED) == 0) { error = EPERM; goto fail; } /* not being traced by YOU */ if (p->p_pptr != td->td_proc) { error = EBUSY; goto fail; } /* not currently stopped */ if (!P_SHOULDSTOP(p) || (p->p_flag & P_WAITED) == 0) { error = EBUSY; goto fail; } if (req == PTRACE_GETFPXREGS) { _PHOLD(p); /* may block */ td2 = FIRST_THREAD_IN_PROC(p); error = linux_proc_read_fpxregs(td2, &r.fpxreg); _PRELE(p); PROC_UNLOCK(p); if (error == 0) error = copyout(&r.fpxreg, (void *)uap->data, sizeof(r.fpxreg)); } else { /* clear dangerous bits exactly as Linux does*/ r.fpxreg.mxcsr &= 0xffbf; _PHOLD(p); /* may block */ td2 = FIRST_THREAD_IN_PROC(p); error = linux_proc_write_fpxregs(td2, &r.fpxreg); _PRELE(p); PROC_UNLOCK(p); } break; fail: PROC_UNLOCK(p); #else error = EIO; #endif break; } case PTRACE_PEEKUSR: case PTRACE_POKEUSR: { error = EIO; /* check addr for alignment */ if (uap->addr < 0 || uap->addr & (sizeof(l_int) - 1)) break; /* * Allow linux programs to access register values in * user struct. We simulate this through PT_GET/SETREGS * as necessary. */ if (uap->addr < sizeof(struct linux_pt_reg)) { error = kern_ptrace(td, PT_GETREGS, pid, &u.bsd_reg, 0); if (error != 0) break; map_regs_to_linux(&u.bsd_reg, &r.reg); if (req == PTRACE_PEEKUSR) { error = copyout((char *)&r.reg + uap->addr, (void *)uap->data, sizeof(l_int)); break; } *(l_int *)((char *)&r.reg + uap->addr) = (l_int)uap->data; map_regs_from_linux(&u.bsd_reg, &r.reg); error = kern_ptrace(td, PT_SETREGS, pid, &u.bsd_reg, 0); } /* * Simulate debug registers access */ if (uap->addr >= LINUX_DBREG_OFFSET && uap->addr <= LINUX_DBREG_OFFSET + LINUX_DBREG_SIZE) { error = kern_ptrace(td, PT_GETDBREGS, pid, &u.bsd_dbreg, 0); if (error != 0) break; uap->addr -= LINUX_DBREG_OFFSET; if (req == PTRACE_PEEKUSR) { error = copyout((char *)&u.bsd_dbreg + uap->addr, (void *)uap->data, sizeof(l_int)); break; } *(l_int *)((char *)&u.bsd_dbreg + uap->addr) = uap->data; error = kern_ptrace(td, PT_SETDBREGS, pid, &u.bsd_dbreg, 0); } break; } case PTRACE_SYSCALL: /* fall through */ default: printf("linux: ptrace(%u, ...) not implemented\n", (unsigned int)uap->req); error = EINVAL; break; } return (error); }
int fork1(struct thread *td, struct fork_req *fr) { struct proc *p1, *newproc; struct thread *td2; struct vmspace *vm2; struct file *fp_procdesc; vm_ooffset_t mem_charged; int error, nprocs_new, ok; static int curfail; static struct timeval lastfail; int flags, pages; flags = fr->fr_flags; pages = fr->fr_pages; if ((flags & RFSTOPPED) != 0) MPASS(fr->fr_procp != NULL && fr->fr_pidp == NULL); else MPASS(fr->fr_procp == NULL); /* Check for the undefined or unimplemented flags. */ if ((flags & ~(RFFLAGS | RFTSIGFLAGS(RFTSIGMASK))) != 0) return (EINVAL); /* Signal value requires RFTSIGZMB. */ if ((flags & RFTSIGFLAGS(RFTSIGMASK)) != 0 && (flags & RFTSIGZMB) == 0) return (EINVAL); /* Can't copy and clear. */ if ((flags & (RFFDG|RFCFDG)) == (RFFDG|RFCFDG)) return (EINVAL); /* Check the validity of the signal number. */ if ((flags & RFTSIGZMB) != 0 && (u_int)RFTSIGNUM(flags) > _SIG_MAXSIG) return (EINVAL); if ((flags & RFPROCDESC) != 0) { /* Can't not create a process yet get a process descriptor. */ if ((flags & RFPROC) == 0) return (EINVAL); /* Must provide a place to put a procdesc if creating one. */ if (fr->fr_pd_fd == NULL) return (EINVAL); /* Check if we are using supported flags. */ if ((fr->fr_pd_flags & ~PD_ALLOWED_AT_FORK) != 0) return (EINVAL); } p1 = td->td_proc; /* * Here we don't create a new process, but we divorce * certain parts of a process from itself. */ if ((flags & RFPROC) == 0) { if (fr->fr_procp != NULL) *fr->fr_procp = NULL; else if (fr->fr_pidp != NULL) *fr->fr_pidp = 0; return (fork_norfproc(td, flags)); } fp_procdesc = NULL; newproc = NULL; vm2 = NULL; /* * Increment the nprocs resource before allocations occur. * Although process entries are dynamically created, we still * keep a global limit on the maximum number we will * create. There are hard-limits as to the number of processes * that can run, established by the KVA and memory usage for * the process data. * * Don't allow a nonprivileged user to use the last ten * processes; don't let root exceed the limit. */ nprocs_new = atomic_fetchadd_int(&nprocs, 1) + 1; if ((nprocs_new >= maxproc - 10 && priv_check_cred(td->td_ucred, PRIV_MAXPROC, 0) != 0) || nprocs_new >= maxproc) { error = EAGAIN; sx_xlock(&allproc_lock); if (ppsratecheck(&lastfail, &curfail, 1)) { printf("maxproc limit exceeded by uid %u (pid %d); " "see tuning(7) and login.conf(5)\n", td->td_ucred->cr_ruid, p1->p_pid); } sx_xunlock(&allproc_lock); goto fail2; } /* * If required, create a process descriptor in the parent first; we * will abandon it if something goes wrong. We don't finit() until * later. */ if (flags & RFPROCDESC) { error = procdesc_falloc(td, &fp_procdesc, fr->fr_pd_fd, fr->fr_pd_flags, fr->fr_pd_fcaps); if (error != 0) goto fail2; } mem_charged = 0; if (pages == 0) pages = kstack_pages; /* Allocate new proc. */ newproc = uma_zalloc(proc_zone, M_WAITOK); td2 = FIRST_THREAD_IN_PROC(newproc); if (td2 == NULL) { td2 = thread_alloc(pages); if (td2 == NULL) { error = ENOMEM; goto fail2; } proc_linkup(newproc, td2); } else { if (td2->td_kstack == 0 || td2->td_kstack_pages != pages) { if (td2->td_kstack != 0) vm_thread_dispose(td2); if (!thread_alloc_stack(td2, pages)) { error = ENOMEM; goto fail2; } } } if ((flags & RFMEM) == 0) { vm2 = vmspace_fork(p1->p_vmspace, &mem_charged); if (vm2 == NULL) { error = ENOMEM; goto fail2; } if (!swap_reserve(mem_charged)) { /* * The swap reservation failed. The accounting * from the entries of the copied vm2 will be * subtracted in vmspace_free(), so force the * reservation there. */ swap_reserve_force(mem_charged); error = ENOMEM; goto fail2; } } else vm2 = NULL; /* * XXX: This is ugly; when we copy resource usage, we need to bump * per-cred resource counters. */ proc_set_cred_init(newproc, crhold(td->td_ucred)); /* * Initialize resource accounting for the child process. */ error = racct_proc_fork(p1, newproc); if (error != 0) { error = EAGAIN; goto fail1; } #ifdef MAC mac_proc_init(newproc); #endif newproc->p_klist = knlist_alloc(&newproc->p_mtx); STAILQ_INIT(&newproc->p_ktr); /* We have to lock the process tree while we look for a pid. */ sx_slock(&proctree_lock); sx_xlock(&allproc_lock); /* * Increment the count of procs running with this uid. Don't allow * a nonprivileged user to exceed their current limit. * * XXXRW: Can we avoid privilege here if it's not needed? */ error = priv_check_cred(td->td_ucred, PRIV_PROC_LIMIT, 0); if (error == 0) ok = chgproccnt(td->td_ucred->cr_ruidinfo, 1, 0); else { ok = chgproccnt(td->td_ucred->cr_ruidinfo, 1, lim_cur(td, RLIMIT_NPROC)); } if (ok) { do_fork(td, fr, newproc, td2, vm2, fp_procdesc); return (0); } error = EAGAIN; sx_sunlock(&proctree_lock); sx_xunlock(&allproc_lock); #ifdef MAC mac_proc_destroy(newproc); #endif racct_proc_exit(newproc); fail1: crfree(newproc->p_ucred); newproc->p_ucred = NULL; fail2: if (vm2 != NULL) vmspace_free(vm2); uma_zfree(proc_zone, newproc); if ((flags & RFPROCDESC) != 0 && fp_procdesc != NULL) { fdclose(td, fp_procdesc, *fr->fr_pd_fd); fdrop(fp_procdesc, td); } atomic_add_int(&nprocs, -1); pause("fork", hz / 2); return (error); }
static int procfs_control(struct thread *td, struct proc *p, int op) { int error = 0; struct thread *temp; /* * Attach - attaches the target process for debugging * by the calling process. */ if (op == PROCFS_CTL_ATTACH) { sx_xlock(&proctree_lock); PROC_LOCK(p); if ((error = p_candebug(td, p)) != 0) goto out; if (p->p_flag & P_TRACED) { error = EBUSY; goto out; } /* Can't trace yourself! */ if (p->p_pid == td->td_proc->p_pid) { error = EINVAL; goto out; } /* * Go ahead and set the trace flag. * Save the old parent (it's reset in * _DETACH, and also in kern_exit.c:wait4() * Reparent the process so that the tracing * proc gets to see all the action. * Stop the target. */ p->p_flag |= P_TRACED; faultin(p); p->p_xstat = 0; /* XXX ? */ if (p->p_pptr != td->td_proc) { p->p_oppid = p->p_pptr->p_pid; proc_reparent(p, td->td_proc); } psignal(p, SIGSTOP); out: PROC_UNLOCK(p); sx_xunlock(&proctree_lock); return (error); } /* * Authorization check: rely on normal debugging protection, except * allow processes to disengage debugging on a process onto which * they have previously attached, but no longer have permission to * debug. */ PROC_LOCK(p); if (op != PROCFS_CTL_DETACH && ((error = p_candebug(td, p)))) { PROC_UNLOCK(p); return (error); } /* * Target process must be stopped, owned by (td) and * be set up for tracing (P_TRACED flag set). * Allow DETACH to take place at any time for sanity. * Allow WAIT any time, of course. */ switch (op) { case PROCFS_CTL_DETACH: case PROCFS_CTL_WAIT: break; default: if (!TRACE_WAIT_P(td->td_proc, p)) { PROC_UNLOCK(p); return (EBUSY); } } #ifdef FIX_SSTEP /* * do single-step fixup if needed */ FIX_SSTEP(FIRST_THREAD_IN_PROC(p)); #endif /* * Don't deliver any signal by default. * To continue with a signal, just send * the signal name to the ctl file */ p->p_xstat = 0; switch (op) { /* * Detach. Cleans up the target process, reparent it if possible * and set it running once more. */ case PROCFS_CTL_DETACH: /* if not being traced, then this is a painless no-op */ if ((p->p_flag & P_TRACED) == 0) { PROC_UNLOCK(p); return (0); } /* not being traced any more */ p->p_flag &= ~(P_TRACED | P_STOPPED_TRACE); /* remove pending SIGTRAP, else the process will die */ sigqueue_delete_proc(p, SIGTRAP); FOREACH_THREAD_IN_PROC(p, temp) temp->td_dbgflags &= ~TDB_SUSPEND; PROC_UNLOCK(p); /* give process back to original parent */ sx_xlock(&proctree_lock); if (p->p_oppid != p->p_pptr->p_pid) { struct proc *pp; pp = pfind(p->p_oppid); PROC_LOCK(p); if (pp) { PROC_UNLOCK(pp); proc_reparent(p, pp); } } else PROC_LOCK(p); p->p_oppid = 0; p->p_flag &= ~P_WAITED; /* XXX ? */ sx_xunlock(&proctree_lock); wakeup(td->td_proc); /* XXX for CTL_WAIT below ? */ break; /* * Step. Let the target process execute a single instruction. * What does it mean to single step a threaded program? */ case PROCFS_CTL_STEP: error = proc_sstep(FIRST_THREAD_IN_PROC(p)); if (error) { PROC_UNLOCK(p); return (error); } break; /* * Run. Let the target process continue running until a breakpoint * or some other trap. */ case PROCFS_CTL_RUN: p->p_flag &= ~P_STOPPED_SIG; /* this uses SIGSTOP */ break; /* * Wait for the target process to stop. * If the target is not being traced then just wait * to enter */ case PROCFS_CTL_WAIT: if (p->p_flag & P_TRACED) { while (error == 0 && (P_SHOULDSTOP(p)) && (p->p_flag & P_TRACED) && (p->p_pptr == td->td_proc)) error = msleep(p, &p->p_mtx, PWAIT|PCATCH, "procfsx", 0); if (error == 0 && !TRACE_WAIT_P(td->td_proc, p)) error = EBUSY; } else { while (error == 0 && P_SHOULDSTOP(p)) error = msleep(p, &p->p_mtx, PWAIT|PCATCH, "procfs", 0); } PROC_UNLOCK(p); return (error); default: panic("procfs_control"); } PROC_SLOCK(p); thread_unsuspend(p); /* If it can run, let it do so. */ PROC_SUNLOCK(p); PROC_UNLOCK(p); return (0); }