Index: head/sys/amd64/vmm/vmm_dev.c =================================================================== --- head/sys/amd64/vmm/vmm_dev.c (revision 337022) +++ head/sys/amd64/vmm/vmm_dev.c (revision 337023) @@ -1,1093 +1,1131 @@ /*- * SPDX-License-Identifier: BSD-2-Clause-FreeBSD * * Copyright (c) 2011 NetApp, Inc. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY NETAPP, INC ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL NETAPP, INC OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include __FBSDID("$FreeBSD$"); #include #include +#include #include #include #include #include #include #include #include #include #include #include +#include #include #include #include #include #include #include #include #include #include "vmm_lapic.h" #include "vmm_stat.h" #include "vmm_mem.h" #include "io/ppt.h" #include "io/vatpic.h" #include "io/vioapic.h" #include "io/vhpet.h" #include "io/vrtc.h" struct devmem_softc { int segid; char *name; struct cdev *cdev; struct vmmdev_softc *sc; SLIST_ENTRY(devmem_softc) link; }; struct vmmdev_softc { struct vm *vm; /* vm instance cookie */ struct cdev *cdev; SLIST_ENTRY(vmmdev_softc) link; SLIST_HEAD(, devmem_softc) devmem; int flags; }; #define VSC_LINKED 0x01 static SLIST_HEAD(, vmmdev_softc) head; +static unsigned pr_allow_flag; static struct mtx vmmdev_mtx; static MALLOC_DEFINE(M_VMMDEV, "vmmdev", "vmmdev"); SYSCTL_DECL(_hw_vmm); +static int vmm_priv_check(struct ucred *ucred); static int devmem_create_cdev(const char *vmname, int id, char *devmem); static void devmem_destroy(void *arg); static int +vmm_priv_check(struct ucred *ucred) +{ + + if (jailed(ucred) && + !(ucred->cr_prison->pr_allow & pr_allow_flag)) + return (EPERM); + + return (0); +} + +static int vcpu_lock_one(struct vmmdev_softc *sc, int vcpu) { int error; if (vcpu < 0 || vcpu >= VM_MAXCPU) return (EINVAL); error = vcpu_set_state(sc->vm, vcpu, VCPU_FROZEN, true); return (error); } static void vcpu_unlock_one(struct vmmdev_softc *sc, int vcpu) { enum vcpu_state state; state = vcpu_get_state(sc->vm, vcpu, NULL); if (state != VCPU_FROZEN) { panic("vcpu %s(%d) has invalid state %d", vm_name(sc->vm), vcpu, state); } vcpu_set_state(sc->vm, vcpu, VCPU_IDLE, false); } static int vcpu_lock_all(struct vmmdev_softc *sc) { int error, vcpu; for (vcpu = 0; vcpu < VM_MAXCPU; vcpu++) { error = vcpu_lock_one(sc, vcpu); if (error) break; } if (error) { while (--vcpu >= 0) vcpu_unlock_one(sc, vcpu); } return (error); } static void vcpu_unlock_all(struct vmmdev_softc *sc) { int vcpu; for (vcpu = 0; vcpu < VM_MAXCPU; vcpu++) vcpu_unlock_one(sc, vcpu); } static struct vmmdev_softc * vmmdev_lookup(const char *name) { struct vmmdev_softc *sc; #ifdef notyet /* XXX kernel is not compiled with invariants */ mtx_assert(&vmmdev_mtx, MA_OWNED); #endif SLIST_FOREACH(sc, &head, link) { if (strcmp(name, vm_name(sc->vm)) == 0) break; } return (sc); } static struct vmmdev_softc * vmmdev_lookup2(struct cdev *cdev) { return (cdev->si_drv1); } static int vmmdev_rw(struct cdev *cdev, struct uio *uio, int flags) { int error, off, c, prot; vm_paddr_t gpa, maxaddr; void *hpa, *cookie; struct vmmdev_softc *sc; + error = vmm_priv_check(curthread->td_ucred); + if (error) + return (error); + sc = vmmdev_lookup2(cdev); if (sc == NULL) return (ENXIO); /* * Get a read lock on the guest memory map by freezing any vcpu. */ error = vcpu_lock_one(sc, VM_MAXCPU - 1); if (error) return (error); prot = (uio->uio_rw == UIO_WRITE ? VM_PROT_WRITE : VM_PROT_READ); maxaddr = vmm_sysmem_maxaddr(sc->vm); while (uio->uio_resid > 0 && error == 0) { gpa = uio->uio_offset; off = gpa & PAGE_MASK; c = min(uio->uio_resid, PAGE_SIZE - off); /* * The VM has a hole in its physical memory map. If we want to * use 'dd' to inspect memory beyond the hole we need to * provide bogus data for memory that lies in the hole. * * Since this device does not support lseek(2), dd(1) will * read(2) blocks of data to simulate the lseek(2). */ hpa = vm_gpa_hold(sc->vm, VM_MAXCPU - 1, gpa, c, prot, &cookie); if (hpa == NULL) { if (uio->uio_rw == UIO_READ && gpa < maxaddr) error = uiomove(__DECONST(void *, zero_region), c, uio); else error = EFAULT; } else { error = uiomove(hpa, c, uio); vm_gpa_release(cookie); } } vcpu_unlock_one(sc, VM_MAXCPU - 1); return (error); } CTASSERT(sizeof(((struct vm_memseg *)0)->name) >= SPECNAMELEN + 1); static int get_memseg(struct vmmdev_softc *sc, struct vm_memseg *mseg) { struct devmem_softc *dsc; int error; bool sysmem; error = vm_get_memseg(sc->vm, mseg->segid, &mseg->len, &sysmem, NULL); if (error || mseg->len == 0) return (error); if (!sysmem) { SLIST_FOREACH(dsc, &sc->devmem, link) { if (dsc->segid == mseg->segid) break; } KASSERT(dsc != NULL, ("%s: devmem segment %d not found", __func__, mseg->segid)); error = copystr(dsc->name, mseg->name, SPECNAMELEN + 1, NULL); } else { bzero(mseg->name, sizeof(mseg->name)); } return (error); } static int alloc_memseg(struct vmmdev_softc *sc, struct vm_memseg *mseg) { char *name; int error; bool sysmem; error = 0; name = NULL; sysmem = true; if (VM_MEMSEG_NAME(mseg)) { sysmem = false; name = malloc(SPECNAMELEN + 1, M_VMMDEV, M_WAITOK); error = copystr(mseg->name, name, SPECNAMELEN + 1, 0); if (error) goto done; } error = vm_alloc_memseg(sc->vm, mseg->segid, mseg->len, sysmem); if (error) goto done; if (VM_MEMSEG_NAME(mseg)) { error = devmem_create_cdev(vm_name(sc->vm), mseg->segid, name); if (error) vm_free_memseg(sc->vm, mseg->segid); else name = NULL; /* freed when 'cdev' is destroyed */ } done: free(name, M_VMMDEV); return (error); } static int vm_get_register_set(struct vm *vm, int vcpu, unsigned int count, int *regnum, uint64_t *regval) { int error, i; error = 0; for (i = 0; i < count; i++) { error = vm_get_register(vm, vcpu, regnum[i], ®val[i]); if (error) break; } return (error); } static int vm_set_register_set(struct vm *vm, int vcpu, unsigned int count, int *regnum, uint64_t *regval) { int error, i; error = 0; for (i = 0; i < count; i++) { error = vm_set_register(vm, vcpu, regnum[i], regval[i]); if (error) break; } return (error); } static int vmmdev_ioctl(struct cdev *cdev, u_long cmd, caddr_t data, int fflag, struct thread *td) { int error, vcpu, state_changed, size; cpuset_t *cpuset; struct vmmdev_softc *sc; struct vm_register *vmreg; struct vm_seg_desc *vmsegdesc; struct vm_register_set *vmregset; struct vm_run *vmrun; struct vm_exception *vmexc; struct vm_lapic_irq *vmirq; struct vm_lapic_msi *vmmsi; struct vm_ioapic_irq *ioapic_irq; struct vm_isa_irq *isa_irq; struct vm_isa_irq_trigger *isa_irq_trigger; struct vm_capability *vmcap; struct vm_pptdev *pptdev; struct vm_pptdev_mmio *pptmmio; struct vm_pptdev_msi *pptmsi; struct vm_pptdev_msix *pptmsix; struct vm_nmi *vmnmi; struct vm_stats *vmstats; struct vm_stat_desc *statdesc; struct vm_x2apic *x2apic; struct vm_gpa_pte *gpapte; struct vm_suspend *vmsuspend; struct vm_gla2gpa *gg; struct vm_activate_cpu *vac; struct vm_cpuset *vm_cpuset; struct vm_intinfo *vmii; struct vm_rtc_time *rtctime; struct vm_rtc_data *rtcdata; struct vm_memmap *mm; struct vm_cpu_topology *topology; uint64_t *regvals; int *regnums; + error = vmm_priv_check(curthread->td_ucred); + if (error) + return (error); + sc = vmmdev_lookup2(cdev); if (sc == NULL) return (ENXIO); - error = 0; vcpu = -1; state_changed = 0; /* * Some VMM ioctls can operate only on vcpus that are not running. */ switch (cmd) { case VM_RUN: case VM_GET_REGISTER: case VM_SET_REGISTER: case VM_GET_SEGMENT_DESCRIPTOR: case VM_SET_SEGMENT_DESCRIPTOR: case VM_GET_REGISTER_SET: case VM_SET_REGISTER_SET: case VM_INJECT_EXCEPTION: case VM_GET_CAPABILITY: case VM_SET_CAPABILITY: case VM_PPTDEV_MSI: case VM_PPTDEV_MSIX: case VM_SET_X2APIC_STATE: case VM_GLA2GPA: case VM_GLA2GPA_NOFAULT: case VM_ACTIVATE_CPU: case VM_SET_INTINFO: case VM_GET_INTINFO: case VM_RESTART_INSTRUCTION: /* * XXX fragile, handle with care * Assumes that the first field of the ioctl data is the vcpu. */ vcpu = *(int *)data; error = vcpu_lock_one(sc, vcpu); if (error) goto done; state_changed = 1; break; case VM_MAP_PPTDEV_MMIO: case VM_BIND_PPTDEV: case VM_UNBIND_PPTDEV: case VM_ALLOC_MEMSEG: case VM_MMAP_MEMSEG: case VM_REINIT: /* * ioctls that operate on the entire virtual machine must * prevent all vcpus from running. */ error = vcpu_lock_all(sc); if (error) goto done; state_changed = 2; break; case VM_GET_MEMSEG: case VM_MMAP_GETNEXT: /* * Lock a vcpu to make sure that the memory map cannot be * modified while it is being inspected. */ vcpu = VM_MAXCPU - 1; error = vcpu_lock_one(sc, vcpu); if (error) goto done; state_changed = 1; break; default: break; } switch(cmd) { case VM_RUN: vmrun = (struct vm_run *)data; error = vm_run(sc->vm, vmrun); break; case VM_SUSPEND: vmsuspend = (struct vm_suspend *)data; error = vm_suspend(sc->vm, vmsuspend->how); break; case VM_REINIT: error = vm_reinit(sc->vm); break; case VM_STAT_DESC: { statdesc = (struct vm_stat_desc *)data; error = vmm_stat_desc_copy(statdesc->index, statdesc->desc, sizeof(statdesc->desc)); break; } case VM_STATS: { CTASSERT(MAX_VM_STATS >= MAX_VMM_STAT_ELEMS); vmstats = (struct vm_stats *)data; getmicrotime(&vmstats->tv); error = vmm_stat_copy(sc->vm, vmstats->cpuid, &vmstats->num_entries, vmstats->statbuf); break; } case VM_PPTDEV_MSI: pptmsi = (struct vm_pptdev_msi *)data; error = ppt_setup_msi(sc->vm, pptmsi->vcpu, pptmsi->bus, pptmsi->slot, pptmsi->func, pptmsi->addr, pptmsi->msg, pptmsi->numvec); break; case VM_PPTDEV_MSIX: pptmsix = (struct vm_pptdev_msix *)data; error = ppt_setup_msix(sc->vm, pptmsix->vcpu, pptmsix->bus, pptmsix->slot, pptmsix->func, pptmsix->idx, pptmsix->addr, pptmsix->msg, pptmsix->vector_control); break; case VM_MAP_PPTDEV_MMIO: pptmmio = (struct vm_pptdev_mmio *)data; error = ppt_map_mmio(sc->vm, pptmmio->bus, pptmmio->slot, pptmmio->func, pptmmio->gpa, pptmmio->len, pptmmio->hpa); break; case VM_BIND_PPTDEV: pptdev = (struct vm_pptdev *)data; error = vm_assign_pptdev(sc->vm, pptdev->bus, pptdev->slot, pptdev->func); break; case VM_UNBIND_PPTDEV: pptdev = (struct vm_pptdev *)data; error = vm_unassign_pptdev(sc->vm, pptdev->bus, pptdev->slot, pptdev->func); break; case VM_INJECT_EXCEPTION: vmexc = (struct vm_exception *)data; error = vm_inject_exception(sc->vm, vmexc->cpuid, vmexc->vector, vmexc->error_code_valid, vmexc->error_code, vmexc->restart_instruction); break; case VM_INJECT_NMI: vmnmi = (struct vm_nmi *)data; error = vm_inject_nmi(sc->vm, vmnmi->cpuid); break; case VM_LAPIC_IRQ: vmirq = (struct vm_lapic_irq *)data; error = lapic_intr_edge(sc->vm, vmirq->cpuid, vmirq->vector); break; case VM_LAPIC_LOCAL_IRQ: vmirq = (struct vm_lapic_irq *)data; error = lapic_set_local_intr(sc->vm, vmirq->cpuid, vmirq->vector); break; case VM_LAPIC_MSI: vmmsi = (struct vm_lapic_msi *)data; error = lapic_intr_msi(sc->vm, vmmsi->addr, vmmsi->msg); break; case VM_IOAPIC_ASSERT_IRQ: ioapic_irq = (struct vm_ioapic_irq *)data; error = vioapic_assert_irq(sc->vm, ioapic_irq->irq); break; case VM_IOAPIC_DEASSERT_IRQ: ioapic_irq = (struct vm_ioapic_irq *)data; error = vioapic_deassert_irq(sc->vm, ioapic_irq->irq); break; case VM_IOAPIC_PULSE_IRQ: ioapic_irq = (struct vm_ioapic_irq *)data; error = vioapic_pulse_irq(sc->vm, ioapic_irq->irq); break; case VM_IOAPIC_PINCOUNT: *(int *)data = vioapic_pincount(sc->vm); break; case VM_ISA_ASSERT_IRQ: isa_irq = (struct vm_isa_irq *)data; error = vatpic_assert_irq(sc->vm, isa_irq->atpic_irq); if (error == 0 && isa_irq->ioapic_irq != -1) error = vioapic_assert_irq(sc->vm, isa_irq->ioapic_irq); break; case VM_ISA_DEASSERT_IRQ: isa_irq = (struct vm_isa_irq *)data; error = vatpic_deassert_irq(sc->vm, isa_irq->atpic_irq); if (error == 0 && isa_irq->ioapic_irq != -1) error = vioapic_deassert_irq(sc->vm, isa_irq->ioapic_irq); break; case VM_ISA_PULSE_IRQ: isa_irq = (struct vm_isa_irq *)data; error = vatpic_pulse_irq(sc->vm, isa_irq->atpic_irq); if (error == 0 && isa_irq->ioapic_irq != -1) error = vioapic_pulse_irq(sc->vm, isa_irq->ioapic_irq); break; case VM_ISA_SET_IRQ_TRIGGER: isa_irq_trigger = (struct vm_isa_irq_trigger *)data; error = vatpic_set_irq_trigger(sc->vm, isa_irq_trigger->atpic_irq, isa_irq_trigger->trigger); break; case VM_MMAP_GETNEXT: mm = (struct vm_memmap *)data; error = vm_mmap_getnext(sc->vm, &mm->gpa, &mm->segid, &mm->segoff, &mm->len, &mm->prot, &mm->flags); break; case VM_MMAP_MEMSEG: mm = (struct vm_memmap *)data; error = vm_mmap_memseg(sc->vm, mm->gpa, mm->segid, mm->segoff, mm->len, mm->prot, mm->flags); break; case VM_ALLOC_MEMSEG: error = alloc_memseg(sc, (struct vm_memseg *)data); break; case VM_GET_MEMSEG: error = get_memseg(sc, (struct vm_memseg *)data); break; case VM_GET_REGISTER: vmreg = (struct vm_register *)data; error = vm_get_register(sc->vm, vmreg->cpuid, vmreg->regnum, &vmreg->regval); break; case VM_SET_REGISTER: vmreg = (struct vm_register *)data; error = vm_set_register(sc->vm, vmreg->cpuid, vmreg->regnum, vmreg->regval); break; case VM_SET_SEGMENT_DESCRIPTOR: vmsegdesc = (struct vm_seg_desc *)data; error = vm_set_seg_desc(sc->vm, vmsegdesc->cpuid, vmsegdesc->regnum, &vmsegdesc->desc); break; case VM_GET_SEGMENT_DESCRIPTOR: vmsegdesc = (struct vm_seg_desc *)data; error = vm_get_seg_desc(sc->vm, vmsegdesc->cpuid, vmsegdesc->regnum, &vmsegdesc->desc); break; case VM_GET_REGISTER_SET: vmregset = (struct vm_register_set *)data; if (vmregset->count > VM_REG_LAST) { error = EINVAL; break; } regvals = malloc(sizeof(regvals[0]) * vmregset->count, M_VMMDEV, M_WAITOK); regnums = malloc(sizeof(regnums[0]) * vmregset->count, M_VMMDEV, M_WAITOK); error = copyin(vmregset->regnums, regnums, sizeof(regnums[0]) * vmregset->count); if (error == 0) error = vm_get_register_set(sc->vm, vmregset->cpuid, vmregset->count, regnums, regvals); if (error == 0) error = copyout(regvals, vmregset->regvals, sizeof(regvals[0]) * vmregset->count); free(regvals, M_VMMDEV); free(regnums, M_VMMDEV); break; case VM_SET_REGISTER_SET: vmregset = (struct vm_register_set *)data; if (vmregset->count > VM_REG_LAST) { error = EINVAL; break; } regvals = malloc(sizeof(regvals[0]) * vmregset->count, M_VMMDEV, M_WAITOK); regnums = malloc(sizeof(regnums[0]) * vmregset->count, M_VMMDEV, M_WAITOK); error = copyin(vmregset->regnums, regnums, sizeof(regnums[0]) * vmregset->count); if (error == 0) error = copyin(vmregset->regvals, regvals, sizeof(regvals[0]) * vmregset->count); if (error == 0) error = vm_set_register_set(sc->vm, vmregset->cpuid, vmregset->count, regnums, regvals); free(regvals, M_VMMDEV); free(regnums, M_VMMDEV); break; case VM_GET_CAPABILITY: vmcap = (struct vm_capability *)data; error = vm_get_capability(sc->vm, vmcap->cpuid, vmcap->captype, &vmcap->capval); break; case VM_SET_CAPABILITY: vmcap = (struct vm_capability *)data; error = vm_set_capability(sc->vm, vmcap->cpuid, vmcap->captype, vmcap->capval); break; case VM_SET_X2APIC_STATE: x2apic = (struct vm_x2apic *)data; error = vm_set_x2apic_state(sc->vm, x2apic->cpuid, x2apic->state); break; case VM_GET_X2APIC_STATE: x2apic = (struct vm_x2apic *)data; error = vm_get_x2apic_state(sc->vm, x2apic->cpuid, &x2apic->state); break; case VM_GET_GPA_PMAP: gpapte = (struct vm_gpa_pte *)data; pmap_get_mapping(vmspace_pmap(vm_get_vmspace(sc->vm)), gpapte->gpa, gpapte->pte, &gpapte->ptenum); error = 0; break; case VM_GET_HPET_CAPABILITIES: error = vhpet_getcap((struct vm_hpet_cap *)data); break; case VM_GLA2GPA: { CTASSERT(PROT_READ == VM_PROT_READ); CTASSERT(PROT_WRITE == VM_PROT_WRITE); CTASSERT(PROT_EXEC == VM_PROT_EXECUTE); gg = (struct vm_gla2gpa *)data; error = vm_gla2gpa(sc->vm, gg->vcpuid, &gg->paging, gg->gla, gg->prot, &gg->gpa, &gg->fault); KASSERT(error == 0 || error == EFAULT, ("%s: vm_gla2gpa unknown error %d", __func__, error)); break; } case VM_GLA2GPA_NOFAULT: gg = (struct vm_gla2gpa *)data; error = vm_gla2gpa_nofault(sc->vm, gg->vcpuid, &gg->paging, gg->gla, gg->prot, &gg->gpa, &gg->fault); KASSERT(error == 0 || error == EFAULT, ("%s: vm_gla2gpa unknown error %d", __func__, error)); break; case VM_ACTIVATE_CPU: vac = (struct vm_activate_cpu *)data; error = vm_activate_cpu(sc->vm, vac->vcpuid); break; case VM_GET_CPUS: error = 0; vm_cpuset = (struct vm_cpuset *)data; size = vm_cpuset->cpusetsize; if (size < sizeof(cpuset_t) || size > CPU_MAXSIZE / NBBY) { error = ERANGE; break; } cpuset = malloc(size, M_TEMP, M_WAITOK | M_ZERO); if (vm_cpuset->which == VM_ACTIVE_CPUS) *cpuset = vm_active_cpus(sc->vm); else if (vm_cpuset->which == VM_SUSPENDED_CPUS) *cpuset = vm_suspended_cpus(sc->vm); else if (vm_cpuset->which == VM_DEBUG_CPUS) *cpuset = vm_debug_cpus(sc->vm); else error = EINVAL; if (error == 0) error = copyout(cpuset, vm_cpuset->cpus, size); free(cpuset, M_TEMP); break; case VM_SUSPEND_CPU: vac = (struct vm_activate_cpu *)data; error = vm_suspend_cpu(sc->vm, vac->vcpuid); break; case VM_RESUME_CPU: vac = (struct vm_activate_cpu *)data; error = vm_resume_cpu(sc->vm, vac->vcpuid); break; case VM_SET_INTINFO: vmii = (struct vm_intinfo *)data; error = vm_exit_intinfo(sc->vm, vmii->vcpuid, vmii->info1); break; case VM_GET_INTINFO: vmii = (struct vm_intinfo *)data; error = vm_get_intinfo(sc->vm, vmii->vcpuid, &vmii->info1, &vmii->info2); break; case VM_RTC_WRITE: rtcdata = (struct vm_rtc_data *)data; error = vrtc_nvram_write(sc->vm, rtcdata->offset, rtcdata->value); break; case VM_RTC_READ: rtcdata = (struct vm_rtc_data *)data; error = vrtc_nvram_read(sc->vm, rtcdata->offset, &rtcdata->value); break; case VM_RTC_SETTIME: rtctime = (struct vm_rtc_time *)data; error = vrtc_set_time(sc->vm, rtctime->secs); break; case VM_RTC_GETTIME: error = 0; rtctime = (struct vm_rtc_time *)data; rtctime->secs = vrtc_get_time(sc->vm); break; case VM_RESTART_INSTRUCTION: error = vm_restart_instruction(sc->vm, vcpu); break; case VM_SET_TOPOLOGY: topology = (struct vm_cpu_topology *)data; error = vm_set_topology(sc->vm, topology->sockets, topology->cores, topology->threads, topology->maxcpus); break; case VM_GET_TOPOLOGY: topology = (struct vm_cpu_topology *)data; vm_get_topology(sc->vm, &topology->sockets, &topology->cores, &topology->threads, &topology->maxcpus); error = 0; break; default: error = ENOTTY; break; } if (state_changed == 1) vcpu_unlock_one(sc, vcpu); else if (state_changed == 2) vcpu_unlock_all(sc); done: /* Make sure that no handler returns a bogus value like ERESTART */ KASSERT(error >= 0, ("vmmdev_ioctl: invalid error return %d", error)); return (error); } static int vmmdev_mmap_single(struct cdev *cdev, vm_ooffset_t *offset, vm_size_t mapsize, struct vm_object **objp, int nprot) { struct vmmdev_softc *sc; vm_paddr_t gpa; size_t len; vm_ooffset_t segoff, first, last; int error, found, segid; bool sysmem; + error = vmm_priv_check(curthread->td_ucred); + if (error) + return (error); + first = *offset; last = first + mapsize; if ((nprot & PROT_EXEC) || first < 0 || first >= last) return (EINVAL); sc = vmmdev_lookup2(cdev); if (sc == NULL) { /* virtual machine is in the process of being created */ return (EINVAL); } /* * Get a read lock on the guest memory map by freezing any vcpu. */ error = vcpu_lock_one(sc, VM_MAXCPU - 1); if (error) return (error); gpa = 0; found = 0; while (!found) { error = vm_mmap_getnext(sc->vm, &gpa, &segid, &segoff, &len, NULL, NULL); if (error) break; if (first >= gpa && last <= gpa + len) found = 1; else gpa += len; } if (found) { error = vm_get_memseg(sc->vm, segid, &len, &sysmem, objp); KASSERT(error == 0 && *objp != NULL, ("%s: invalid memory segment %d", __func__, segid)); if (sysmem) { vm_object_reference(*objp); *offset = segoff + (first - gpa); } else { error = EINVAL; } } vcpu_unlock_one(sc, VM_MAXCPU - 1); return (error); } static void vmmdev_destroy(void *arg) { struct vmmdev_softc *sc = arg; struct devmem_softc *dsc; int error; error = vcpu_lock_all(sc); KASSERT(error == 0, ("%s: error %d freezing vcpus", __func__, error)); while ((dsc = SLIST_FIRST(&sc->devmem)) != NULL) { KASSERT(dsc->cdev == NULL, ("%s: devmem not free", __func__)); SLIST_REMOVE_HEAD(&sc->devmem, link); free(dsc->name, M_VMMDEV); free(dsc, M_VMMDEV); } if (sc->cdev != NULL) destroy_dev(sc->cdev); if (sc->vm != NULL) vm_destroy(sc->vm); if ((sc->flags & VSC_LINKED) != 0) { mtx_lock(&vmmdev_mtx); SLIST_REMOVE(&head, sc, vmmdev_softc, link); mtx_unlock(&vmmdev_mtx); } free(sc, M_VMMDEV); } static int sysctl_vmm_destroy(SYSCTL_HANDLER_ARGS) { int error; char buf[VM_MAX_NAMELEN]; struct devmem_softc *dsc; struct vmmdev_softc *sc; struct cdev *cdev; + error = vmm_priv_check(req->td->td_ucred); + if (error) + return (error); + strlcpy(buf, "beavis", sizeof(buf)); error = sysctl_handle_string(oidp, buf, sizeof(buf), req); if (error != 0 || req->newptr == NULL) return (error); mtx_lock(&vmmdev_mtx); sc = vmmdev_lookup(buf); if (sc == NULL || sc->cdev == NULL) { mtx_unlock(&vmmdev_mtx); return (EINVAL); } /* * The 'cdev' will be destroyed asynchronously when 'si_threadcount' * goes down to 0 so we should not do it again in the callback. * * Setting 'sc->cdev' to NULL is also used to indicate that the VM * is scheduled for destruction. */ cdev = sc->cdev; sc->cdev = NULL; mtx_unlock(&vmmdev_mtx); /* * Schedule all cdevs to be destroyed: * * - any new operations on the 'cdev' will return an error (ENXIO). * * - when the 'si_threadcount' dwindles down to zero the 'cdev' will * be destroyed and the callback will be invoked in a taskqueue * context. * * - the 'devmem' cdevs are destroyed before the virtual machine 'cdev' */ SLIST_FOREACH(dsc, &sc->devmem, link) { KASSERT(dsc->cdev != NULL, ("devmem cdev already destroyed")); destroy_dev_sched_cb(dsc->cdev, devmem_destroy, dsc); } destroy_dev_sched_cb(cdev, vmmdev_destroy, sc); return (0); } -SYSCTL_PROC(_hw_vmm, OID_AUTO, destroy, CTLTYPE_STRING | CTLFLAG_RW, +SYSCTL_PROC(_hw_vmm, OID_AUTO, destroy, + CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_PRISON, NULL, 0, sysctl_vmm_destroy, "A", NULL); static struct cdevsw vmmdevsw = { .d_name = "vmmdev", .d_version = D_VERSION, .d_ioctl = vmmdev_ioctl, .d_mmap_single = vmmdev_mmap_single, .d_read = vmmdev_rw, .d_write = vmmdev_rw, }; static int sysctl_vmm_create(SYSCTL_HANDLER_ARGS) { int error; struct vm *vm; struct cdev *cdev; struct vmmdev_softc *sc, *sc2; char buf[VM_MAX_NAMELEN]; + error = vmm_priv_check(req->td->td_ucred); + if (error) + return (error); + strlcpy(buf, "beavis", sizeof(buf)); error = sysctl_handle_string(oidp, buf, sizeof(buf), req); if (error != 0 || req->newptr == NULL) return (error); mtx_lock(&vmmdev_mtx); sc = vmmdev_lookup(buf); mtx_unlock(&vmmdev_mtx); if (sc != NULL) return (EEXIST); error = vm_create(buf, &vm); if (error != 0) return (error); sc = malloc(sizeof(struct vmmdev_softc), M_VMMDEV, M_WAITOK | M_ZERO); sc->vm = vm; SLIST_INIT(&sc->devmem); /* * Lookup the name again just in case somebody sneaked in when we * dropped the lock. */ mtx_lock(&vmmdev_mtx); sc2 = vmmdev_lookup(buf); if (sc2 == NULL) { SLIST_INSERT_HEAD(&head, sc, link); sc->flags |= VSC_LINKED; } mtx_unlock(&vmmdev_mtx); if (sc2 != NULL) { vmmdev_destroy(sc); return (EEXIST); } error = make_dev_p(MAKEDEV_CHECKNAME, &cdev, &vmmdevsw, NULL, UID_ROOT, GID_WHEEL, 0600, "vmm/%s", buf); if (error != 0) { vmmdev_destroy(sc); return (error); } mtx_lock(&vmmdev_mtx); sc->cdev = cdev; sc->cdev->si_drv1 = sc; mtx_unlock(&vmmdev_mtx); return (0); } -SYSCTL_PROC(_hw_vmm, OID_AUTO, create, CTLTYPE_STRING | CTLFLAG_RW, +SYSCTL_PROC(_hw_vmm, OID_AUTO, create, + CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_PRISON, NULL, 0, sysctl_vmm_create, "A", NULL); void vmmdev_init(void) { mtx_init(&vmmdev_mtx, "vmm device mutex", NULL, MTX_DEF); + pr_allow_flag = prison_add_allow(NULL, "vmm", NULL, + "Allow use of vmm in a jail."); } int vmmdev_cleanup(void) { int error; if (SLIST_EMPTY(&head)) error = 0; else error = EBUSY; return (error); } static int devmem_mmap_single(struct cdev *cdev, vm_ooffset_t *offset, vm_size_t len, struct vm_object **objp, int nprot) { struct devmem_softc *dsc; vm_ooffset_t first, last; size_t seglen; int error; bool sysmem; dsc = cdev->si_drv1; if (dsc == NULL) { /* 'cdev' has been created but is not ready for use */ return (ENXIO); } first = *offset; last = *offset + len; if ((nprot & PROT_EXEC) || first < 0 || first >= last) return (EINVAL); error = vcpu_lock_one(dsc->sc, VM_MAXCPU - 1); if (error) return (error); error = vm_get_memseg(dsc->sc->vm, dsc->segid, &seglen, &sysmem, objp); KASSERT(error == 0 && !sysmem && *objp != NULL, ("%s: invalid devmem segment %d", __func__, dsc->segid)); vcpu_unlock_one(dsc->sc, VM_MAXCPU - 1); if (seglen >= last) { vm_object_reference(*objp); return (0); } else { return (EINVAL); } } static struct cdevsw devmemsw = { .d_name = "devmem", .d_version = D_VERSION, .d_mmap_single = devmem_mmap_single, }; static int devmem_create_cdev(const char *vmname, int segid, char *devname) { struct devmem_softc *dsc; struct vmmdev_softc *sc; struct cdev *cdev; int error; error = make_dev_p(MAKEDEV_CHECKNAME, &cdev, &devmemsw, NULL, UID_ROOT, GID_WHEEL, 0600, "vmm.io/%s.%s", vmname, devname); if (error) return (error); dsc = malloc(sizeof(struct devmem_softc), M_VMMDEV, M_WAITOK | M_ZERO); mtx_lock(&vmmdev_mtx); sc = vmmdev_lookup(vmname); KASSERT(sc != NULL, ("%s: vm %s softc not found", __func__, vmname)); if (sc->cdev == NULL) { /* virtual machine is being created or destroyed */ mtx_unlock(&vmmdev_mtx); free(dsc, M_VMMDEV); destroy_dev_sched_cb(cdev, NULL, 0); return (ENODEV); } dsc->segid = segid; dsc->name = devname; dsc->cdev = cdev; dsc->sc = sc; SLIST_INSERT_HEAD(&sc->devmem, dsc, link); mtx_unlock(&vmmdev_mtx); /* The 'cdev' is ready for use after 'si_drv1' is initialized */ cdev->si_drv1 = dsc; return (0); } static void devmem_destroy(void *arg) { struct devmem_softc *dsc = arg; KASSERT(dsc->cdev, ("%s: devmem cdev already destroyed", __func__)); dsc->cdev = NULL; dsc->sc = NULL; } Index: head/usr.sbin/jail/jail.8 =================================================================== --- head/usr.sbin/jail/jail.8 (revision 337022) +++ head/usr.sbin/jail/jail.8 (revision 337023) @@ -1,1376 +1,1383 @@ .\" Copyright (c) 2000, 2003 Robert N. M. Watson .\" Copyright (c) 2008-2012 James Gritton .\" All rights reserved. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice, this list of conditions and the following disclaimer in the .\" documentation and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF .\" SUCH DAMAGE. .\" .\" $FreeBSD$ .\" -.Dd July 29, 2018 +.Dd July 30, 2018 .Dt JAIL 8 .Os .Sh NAME .Nm jail .Nd "manage system jails" .Sh SYNOPSIS .Nm .Op Fl dhilqv .Op Fl J Ar jid_file .Op Fl u Ar username .Op Fl U Ar username .Op Fl cmr .Ar param Ns = Ns Ar value ... .Op Cm command Ns = Ns Ar command ... .Nm .Op Fl dqv .Op Fl f Ar conf_file .Op Fl p Ar limit .Op Fl cmr .Op Ar jail .Nm .Op Fl qv .Op Fl f Ar conf_file .Op Fl rR .Op Cm * | Ar jail ... .Nm .Op Fl dhilqv .Op Fl J Ar jid_file .Op Fl u Ar username .Op Fl U Ar username .Op Fl n Ar jailname .Op Fl s Ar securelevel .Op Ar path hostname [ Ar ip Ns [ Ns Ar ,... Ns ]] Ar command ... .Sh DESCRIPTION The .Nm utility creates new jails, or modifies or removes existing jails. A jail .Pq or Dq prison is specified via parameters on the command line, or in the .Xr jail.conf 5 file. .Pp At least one of the options .Fl c , .Fl m or .Fl r must be specified. These options are used alone or in combination to describe the operation to perform: .Bl -tag -width indent .It Fl c Create a new jail. The jail .Va jid and .Va name parameters (if specified on the command line) must not refer to an existing jail. .It Fl m Modify an existing jail. One of the .Va jid or .Va name parameters must exist and refer to an existing jail. Some parameters may not be changed on a running jail. .It Fl r Remove the .Ar jail specified by jid or name. All jailed processes are killed, and all jails that are children of this jail are also removed. .It Fl rc Restart an existing jail. The jail is first removed and then re-created, as if .Dq Nm Fl r and .Dq Nm Fl c were run in succession. .It Fl cm Create a jail if it does not exist, or modify the jail if it does exist. .It Fl mr Modify an existing jail. The jail may be restarted if necessary to modify parameters than could not otherwise be changed. .It Fl cmr Create a jail if it doesn't exist, or modify (and possibly restart) the jail if it does exist. .El .Pp Other available options are: .Bl -tag -width indent .It Fl d Allow making changes to a dying jail, equivalent to the .Va allow.dying parameter. .It Fl f Ar conf_file Use configuration file .Ar conf_file instead of the default .Pa /etc/jail.conf . .It Fl h Resolve the .Va host.hostname parameter (or .Va hostname ) and add all IP addresses returned by the resolver to the list of addresses for this jail. This is equivalent to the .Va ip_hostname parameter. .It Fl i Output (only) the jail identifier of the newly created jail(s). This implies the .Fl q option. .It Fl J Ar jid_file Write a .Ar jid_file file, containing the parameters used to start the jail. .It Fl l Run commands in a clean environment. This is deprecated and is equivalent to the exec.clean parameter. .It Fl n Ar jailname Set the jail's name. This is deprecated and is equivalent to the .Va name parameter. .It Fl p Ar limit Limit the number of commands from .Va exec.* that can run simultaneously. .It Fl q Suppress the message printed whenever a jail is created, modified or removed. Only error messages will be printed. .It Fl R A variation of the .Fl r option that removes an existing jail without using the configuration file. No removal-related parameters for this jail will be used \(em the jail will simply be removed. .It Fl s Ar securelevel Set the .Va kern.securelevel MIB entry to the specified value inside the newly created jail. This is deprecated and is equivalent to the .Va securelevel parameter. .It Fl u Ar username The user name from host environment as whom jailed commands should run. This is deprecated and is equivalent to the .Va exec.jail_user and .Va exec.system_jail_user parameters. .It Fl U Ar username The user name from the jailed environment as whom jailed commands should run. This is deprecated and is equivalent to the .Va exec.jail_user parameter. .It Fl v Print a message on every operation, such as running commands and mounting filesystems. .El .Pp If no arguments are given after the options, the operation (except remove) will be performed on all jails specified in the .Xr jail.conf 5 file. A single argument of a jail name will operate only on the specified jail. The .Fl r and .Fl R options can also remove running jails that aren't in the .Xr jail.conf 5 file, specified by name or jid. .Pp An argument of .Dq * is a wildcard that will operate on all jails, regardless of whether they appear in .Xr jail.conf 5 ; this is the surest way for .Fl r to remove all jails. If hierarchical jails exist, a partial-matching wildcard definition may be specified. For example, an argument of .Dq foo.* would apply to jails with names like .Dq foo.bar and .Dq foo.bar.baz . .Pp A jail may be specified with parameters directly on the command line. In this case, the .Xr jail.conf 5 file will not be used. For backward compatibility, the command line may also have four fixed parameters, without names: .Ar path , .Ar hostname , .Ar ip , and .Ar command . This mode will always create a new jail, and the .Fl c and .Fl m options do not apply (and must not be present). .Ss Jail Parameters Parameters in the .Xr jail.conf 5 file, or on the command line, are generally of the form .Dq name=value . Some parameters are boolean, and do not have a value but are set by the name alone with or without a .Dq no prefix, e.g. .Va persist or .Va nopersist . They can also be given the values .Dq true and .Dq false . Other parameters may have more than one value, specified as a comma-separated list or with .Dq += in the configuration file (see .Xr jail.conf 5 for details). .Pp The .Nm utility recognizes two classes of parameters. There are the true jail parameters that are passed to the kernel when the jail is created, which can be seen with .Xr jls 8 , and can (usually) be changed with .Dq Nm Fl m . Then there are pseudo-parameters that are only used by .Nm itself. .Pp Jails have a set of core parameters, and kernel modules can add their own jail parameters. The current set of available parameters can be retrieved via .Dq Nm sysctl Fl d Va security.jail.param . Any parameters not set will be given default values, often based on the current environment. The core parameters are: .Bl -tag -width indent .It Va jid The jail identifier. This will be assigned automatically to a new jail (or can be explicitly set), and can be used to identify the jail for later modification, or for such commands as .Xr jls 8 or .Xr jexec 8 . .It Va name The jail name. This is an arbitrary string that identifies a jail (except it may not contain a .Sq \&. ) . Like the .Va jid , it can be passed to later .Nm commands, or to .Xr jls 8 or .Xr jexec 8 . If no .Va name is supplied, a default is assumed that is the same as the .Va jid . The .Va name parameter is implied by the .Xr jail.conf 5 file format, and need not be explicitly set when using the configuration file. .It Va path The directory which is to be the root of the jail. Any commands run inside the jail, either by .Nm or from .Xr jexec 8 , are run from this directory. .It Va ip4.addr A list of IPv4 addresses assigned to the jail. If this is set, the jail is restricted to using only these addresses. Any attempts to use other addresses fail, and attempts to use wildcard addresses silently use the jailed address instead. For IPv4 the first address given will be used as the source address when source address selection on unbound sockets cannot find a better match. It is only possible to start multiple jails with the same IP address if none of the jails has more than this single overlapping IP address assigned to itself. .It Va ip4.saddrsel A boolean option to change the formerly mentioned behaviour and disable IPv4 source address selection for the jail in favour of the primary IPv4 address of the jail. Source address selection is enabled by default for all jails and the .Va ip4.nosaddrsel setting of a parent jail is not inherited for any child jails. .It Va ip4 Control the availability of IPv4 addresses. Possible values are .Dq inherit to allow unrestricted access to all system addresses, .Dq new to restrict addresses via .Va ip4.addr , and .Dq disable to stop the jail from using IPv4 entirely. Setting the .Va ip4.addr parameter implies a value of .Dq new . .It Va ip6.addr , Va ip6.saddrsel , Va ip6 A set of IPv6 options for the jail, the counterparts to .Va ip4.addr , .Va ip4.saddrsel and .Va ip4 above. .It Va vnet Create the jail with its own virtual network stack, with its own network interfaces, addresses, routing table, etc. The kernel must have been compiled with the .Sy VIMAGE option for this to be available. Possible values are .Dq inherit to use the system network stack, possibly with restricted IP addresses, and .Dq new to create a new network stack. .It Va host.hostname The hostname of the jail. Other similar parameters are .Va host.domainname , .Va host.hostuuid and .Va host.hostid . .It Va host Set the origin of hostname and related information. Possible values are .Dq inherit to use the system information and .Dq new for the jail to use the information from the above fields. Setting any of the above fields implies a value of .Dq new . .It Va securelevel The value of the jail's .Va kern.securelevel sysctl. A jail never has a lower securelevel than its parent system, but by setting this parameter it may have a higher one. If the system securelevel is changed, any jail securelevels will be at least as secure. .It Va devfs_ruleset The number of the devfs ruleset that is enforced for mounting devfs in this jail. A value of zero (default) means no ruleset is enforced. Descendant jails inherit the parent jail's devfs ruleset enforcement. Mounting devfs inside a jail is possible only if the .Va allow.mount and .Va allow.mount.devfs permissions are effective and .Va enforce_statfs is set to a value lower than 2. Devfs rules and rulesets cannot be viewed or modified from inside a jail. .Pp NOTE: It is important that only appropriate device nodes in devfs be exposed to a jail; access to disk devices in the jail may permit processes in the jail to bypass the jail sandboxing by modifying files outside of the jail. See .Xr devfs 8 for information on how to use devfs rules to limit access to entries in the per-jail devfs. A simple devfs ruleset for jails is available as ruleset #4 in .Pa /etc/defaults/devfs.rules . .It Va children.max The number of child jails allowed to be created by this jail (or by other jails under this jail). This limit is zero by default, indicating the jail is not allowed to create child jails. See the .Sx "Hierarchical Jails" section for more information. .It Va children.cur The number of descendants of this jail, including its own child jails and any jails created under them. .It Va enforce_statfs This determines what information processes in a jail are able to get about mount points. It affects the behaviour of the following syscalls: .Xr statfs 2 , .Xr fstatfs 2 , .Xr getfsstat 2 , and .Xr fhstatfs 2 (as well as similar compatibility syscalls). When set to 0, all mount points are available without any restrictions. When set to 1, only mount points below the jail's chroot directory are visible. In addition to that, the path to the jail's chroot directory is removed from the front of their pathnames. When set to 2 (default), above syscalls can operate only on a mount-point where the jail's chroot directory is located. .It Va persist Setting this boolean parameter allows a jail to exist without any processes. Normally, a command is run as part of jail creation, and then the jail is destroyed as its last process exits. A new jail must have either the .Va persist parameter or .Va exec.start or .Va command pseudo-parameter set. .It Va cpuset.id The ID of the cpuset associated with this jail (read-only). .It Va dying This is true if the jail is in the process of shutting down (read-only). .It Va parent The .Va jid of the parent of this jail, or zero if this is a top-level jail (read-only). .It Va osrelease The string for the jail's .Va kern.osrelease sysctl and uname -r. .It Va osreldate The number for the jail's .Va kern.osreldate and uname -K. .It Va allow.* Some restrictions of the jail environment may be set on a per-jail basis. With the exception of .Va allow.set_hostname and .Va allow.reserved_ports , these boolean parameters are off by default. .Bl -tag -width indent .It Va allow.set_hostname The jail's hostname may be changed via .Xr hostname 1 or .Xr sethostname 3 . .It Va allow.sysvipc A process within the jail has access to System V IPC primitives. This is deprecated in favor of the per-module parameters (see below). When this parameter is set, it is equivalent to setting .Va sysvmsg , .Va sysvsem , and .Va sysvshm all to .Dq inherit . .It Va allow.raw_sockets The jail root is allowed to create raw sockets. Setting this parameter allows utilities like .Xr ping 8 and .Xr traceroute 8 to operate inside the jail. If this is set, the source IP addresses are enforced to comply with the IP address bound to the jail, regardless of whether or not the .Dv IP_HDRINCL flag has been set on the socket. Since raw sockets can be used to configure and interact with various network subsystems, extra caution should be used where privileged access to jails is given out to untrusted parties. .It Va allow.chflags Normally, privileged users inside a jail are treated as unprivileged by .Xr chflags 2 . When this parameter is set, such users are treated as privileged, and may manipulate system file flags subject to the usual constraints on .Va kern.securelevel . .It Va allow.mount privileged users inside the jail will be able to mount and unmount file system types marked as jail-friendly. The .Xr lsvfs 1 command can be used to find file system types available for mount from within a jail. This permission is effective only if .Va enforce_statfs is set to a value lower than 2. .It Va allow.mount.devfs privileged users inside the jail will be able to mount and unmount the devfs file system. This permission is effective only together with .Va allow.mount and only when .Va enforce_statfs is set to a value lower than 2. The devfs ruleset should be restricted from the default by using the .Va devfs_ruleset option. .It Va allow.quotas The jail root may administer quotas on the jail's filesystem(s). This includes filesystems that the jail may share with other jails or with non-jailed parts of the system. .It Va allow.socket_af Sockets within a jail are normally restricted to IPv4, IPv6, local (UNIX), and route. This allows access to other protocol stacks that have not had jail functionality added to them. .It Va allow.mlock Locking or unlocking physical pages in memory are normally not available within a jail. When this parameter is set, users may .Xr mlock 2 or .Xr munlock 2 memory subject to .Va security.bsd.unprivileged_mlock and resource limits. .It Va allow.reserved_ports The jail root may bind to ports lower than 1024. .El .El .Pp Kernel modules may add their own parameters, which only exist when the module is loaded. These are typically headed under a parameter named after the module, with values of .Dq inherit to give the jail full use of the module, .Dq new to encapsulate the jail in some module-specific way, and .Dq disable to make the module unavailable to the jail. There also may be other parameters to define jail behavior within the module. Module-specific parameters include: .Bl -tag -width indent .It Va allow.mount.fdescfs privileged users inside the jail will be able to mount and unmount the fdescfs file system. This permission is effective only together with .Va allow.mount and only when .Va enforce_statfs is set to a value lower than 2. .It Va allow.mount.fusefs privileged users inside the jail will be able to mount and unmount fuse-based file systems. This permission is effective only together with .Va allow.mount and only when .Va enforce_statfs is set to a value lower than 2. .It Va allow.mount.nullfs privileged users inside the jail will be able to mount and unmount the nullfs file system. This permission is effective only together with .Va allow.mount and only when .Va enforce_statfs is set to a value lower than 2. .It Va allow.mount.procfs privileged users inside the jail will be able to mount and unmount the procfs file system. This permission is effective only together with .Va allow.mount and only when .Va enforce_statfs is set to a value lower than 2. .It Va allow.mount.linprocfs privileged users inside the jail will be able to mount and unmount the linprocfs file system. This permission is effective only together with .Va allow.mount and only when .Va enforce_statfs is set to a value lower than 2. .It Va allow.mount.linsysfs privileged users inside the jail will be able to mount and unmount the linsysfs file system. This permission is effective only together with .Va allow.mount and only when .Va enforce_statfs is set to a value lower than 2. .It Va allow.mount.tmpfs privileged users inside the jail will be able to mount and unmount the tmpfs file system. This permission is effective only together with .Va allow.mount and only when .Va enforce_statfs is set to a value lower than 2. .It Va allow.mount.zfs privileged users inside the jail will be able to mount and unmount the ZFS file system. This permission is effective only together with .Va allow.mount and only when .Va enforce_statfs is set to a value lower than 2. See .Xr zfs 8 for information on how to configure the ZFS filesystem to operate from within a jail. +.It Va allow.vmm +The jail may access +.Xr vmm 4 . +This flag is only available when the +.Xr vmm 4 +kernel module is loaded. .It Va linux Determine how a jail's Linux emulation environment appears. A value of .Dq inherit will keep the same environment, and .Dq new will give the jail it's own environment (still originally inherited when the jail is created). .It Va linux.osname , linux.osrelease , linux.oss_version The Linux OS name, OS release, and OSS version associated with this jail. .It Va sysvmsg Allow access to SYSV IPC message primitives. If set to .Dq inherit , all IPC objects on the system are visible to this jail, whether they were created by the jail itself, the base system, or other jails. If set to .Dq new , the jail will have its own key namespace, and can only see the objects that it has created; the system (or parent jail) has access to the jail's objects, but not to its keys. If set to .Dq disable , the jail cannot perform any sysvmsg-related system calls. .It Va sysvsem, sysvshm Allow access to SYSV IPC semaphore and shared memory primitives, in the same manner as .Va sysvmsg. .El .Pp There are pseudo-parameters that are not passed to the kernel, but are used by .Nm to set up the jail environment, often by running specified commands when jails are created or removed. The .Va exec.* command parameters are .Xr sh 1 command lines that are run in either the system or jail environment. They may be given multiple values, which would run the specified commands in sequence. All commands must succeed (return a zero exit status), or the jail will not be created or removed, as appropriate. .Pp The pseudo-parameters are: .Bl -tag -width indent .It Va exec.prestart Command(s) to run in the system environment before a jail is created. .It Va exec.start Command(s) to run in the jail environment when a jail is created. A typical command to run is .Dq sh /etc/rc . .It Va command A synonym for .Va exec.start for use when specifying a jail directly on the command line. Unlike other parameters whose value is a single string, .Va command uses the remainder of the .Nm command line as its own arguments. .It Va exec.poststart Command(s) to run in the system environment after a jail is created, and after any .Va exec.start commands have completed. .It Va exec.prestop Command(s) to run in the system environment before a jail is removed. .It Va exec.stop Command(s) to run in the jail environment before a jail is removed, and after any .Va exec.prestop commands have completed. A typical command to run is .Dq sh /etc/rc.shutdown . .It Va exec.poststop Command(s) to run in the system environment after a jail is removed. .It Va exec.clean Run commands in a clean environment. The environment is discarded except for .Ev HOME , SHELL , TERM and .Ev USER . .Ev HOME and .Ev SHELL are set to the target login's default values. .Ev USER is set to the target login. .Ev TERM is imported from the current environment. The environment variables from the login class capability database for the target login are also set. .It Va exec.jail_user The user to run commands as, when running in the jail environment. The default is to run the commands as the current user. .It Va exec.system_jail_user This boolean option looks for the .Va exec.jail_user in the system .Xr passwd 5 file, instead of in the jail's file. .It Va exec.system_user The user to run commands as, when running in the system environment. The default is to run the commands as the current user. .It Va exec.timeout The maximum amount of time to wait for a command to complete, in seconds. If a command is still running after this timeout has passed, the jail will not be created or removed, as appropriate. .It Va exec.consolelog A file to direct command output (stdout and stderr) to. .It Va exec.fib The FIB (routing table) to set when running commands inside the jail. .It Va stop.timeout The maximum amount of time to wait for a jail's processes to exit after sending them a .Dv SIGTERM signal (which happens after the .Va exec.stop commands have completed). After this many seconds have passed, the jail will be removed, which will kill any remaining processes. If this is set to zero, no .Dv SIGTERM is sent and the jail is immediately removed. The default is 10 seconds. .It Va interface A network interface to add the jail's IP addresses .Va ( ip4.addr and .Va ip6.addr ) to. An alias for each address will be added to the interface before the jail is created, and will be removed from the interface after the jail is removed. .It Va ip4.addr In addition to the IP addresses that are passed to the kernel, an interface, netmask and additional parameters (as supported by .Xr ifconfig 8 Ns ) may also be specified, in the form .Dq Ar interface Ns | Ns Ar ip-address Ns / Ns Ar netmask param ... . If an interface is given before the IP address, an alias for the address will be added to that interface, as it is with the .Va interface parameter. If a netmask in either dotted-quad or CIDR form is given after an IP address, it will be used when adding the IP alias. If additional parameters are specified then they will also be used when adding the IP alias. .It Va ip6.addr In addition to the IP addresses that are passed to the kernel, an interface, prefix and additional parameters (as supported by .Xr ifconfig 8 Ns ) may also be specified, in the form .Dq Ar interface Ns | Ns Ar ip-address Ns / Ns Ar prefix param ... . .It Va vnet.interface A network interface to give to a vnet-enabled jail after is it created. The interface will automatically be released when the jail is removed. .It Va ip_hostname Resolve the .Va host.hostname parameter and add all IP addresses returned by the resolver to the list of addresses .Po Va ip4.addr or .Va ip6.addr Pc for this jail. This may affect default address selection for outgoing IPv4 connections from jails. The address first returned by the resolver for each address family will be used as the primary address. .It Va mount A filesystem to mount before creating the jail (and to unmount after removing it), given as a single .Xr fstab 5 line. .It Va mount.fstab An .Xr fstab 5 format file containing filesystems to mount before creating a jail. .It Va mount.devfs Mount a .Xr devfs 5 filesystem on the chrooted .Pa /dev directory, and apply the ruleset in the .Va devfs_ruleset parameter (or a default of ruleset 4: devfsrules_jail) to restrict the devices visible inside the jail. .It Va mount.fdescfs Mount a .Xr fdescfs 5 filesystem on the chrooted .Pa /dev/fd directory. .It Va mount.procfs Mount a .Xr procfs 5 filesystem on the chrooted .Pa /proc directory. .It Va allow.dying Allow making changes to a .Va dying jail. .It Va depend Specify a jail (or jails) that this jail depends on. When this jail is to be created, any jail(s) it depends on must already exist. If not, they will be created automatically, up to the completion of the last .Va exec.poststart command, before any action will taken to create this jail. When jails are removed the opposite is true: this jail will be removed, up to the last .Va exec.poststop command, before any jail(s) it depends on are stopped. .El .Sh EXAMPLES Jails are typically set up using one of two philosophies: either to constrain a specific application (possibly running with privilege), or to create a .Dq "virtual system image" running a variety of daemons and services. In both cases, a fairly complete file system install of .Fx is required, so as to provide the necessary command line tools, daemons, libraries, application configuration files, etc. However, for a virtual server configuration, a fair amount of additional work is required so as to replace the .Dq boot process. This manual page documents the configuration steps necessary to support either of these steps, although the configuration steps may need to be refined based on local requirements. .Ss "Setting up a Jail Directory Tree" To set up a jail directory tree containing an entire .Fx distribution, the following .Xr sh 1 command script can be used: .Bd -literal D=/here/is/the/jail cd /usr/src mkdir -p $D make world DESTDIR=$D make distribution DESTDIR=$D .Ed .Pp In many cases this example would put far more in the jail than needed. In the other extreme case a jail might contain only one file: the executable to be run in the jail. .Pp We recommend experimentation, and caution that it is a lot easier to start with a .Dq fat jail and remove things until it stops working, than it is to start with a .Dq thin jail and add things until it works. .Ss "Setting Up a Jail" Do what was described in .Sx "Setting Up a Jail Directory Tree" to build the jail directory tree. For the sake of this example, we will assume you built it in .Pa /data/jail/testjail , for a jail named .Dq testjail . Substitute below as needed with your own directory, IP address, and hostname. .Ss "Setting up the Host Environment" First, set up the real system's environment to be .Dq jail-friendly . For consistency, we will refer to the parent box as the .Dq "host environment" , and to the jailed virtual machine as the .Dq "jail environment" . Since jails are implemented using IP aliases, one of the first things to do is to disable IP services on the host system that listen on all local IP addresses for a service. If a network service is present in the host environment that binds all available IP addresses rather than specific IP addresses, it may service requests sent to jail IP addresses if the jail did not bind the port. This means changing .Xr inetd 8 to only listen on the appropriate IP address, and so forth. Add the following to .Pa /etc/rc.conf in the host environment: .Bd -literal -offset indent sendmail_enable="NO" inetd_flags="-wW -a 192.0.2.23" rpcbind_enable="NO" .Ed .Pp .Li 192.0.2.23 is the native IP address for the host system, in this example. Daemons that run out of .Xr inetd 8 can be easily configured to use only the specified host IP address. Other daemons will need to be manually configured \(em for some this is possible through .Xr rc.conf 5 flags entries; for others it is necessary to modify per-application configuration files, or to recompile the application. The following frequently deployed services must have their individual configuration files modified to limit the application to listening to a specific IP address: .Pp To configure .Xr sshd 8 , it is necessary to modify .Pa /etc/ssh/sshd_config . .Pp To configure .Xr sendmail 8 , it is necessary to modify .Pa /etc/mail/sendmail.cf . .Pp For .Xr named 8 , it is necessary to modify .Pa /etc/namedb/named.conf . .Pp In addition, a number of services must be recompiled in order to run them in the host environment. This includes most applications providing services using .Xr rpc 3 , such as .Xr rpcbind 8 , .Xr nfsd 8 , and .Xr mountd 8 . In general, applications for which it is not possible to specify which IP address to bind should not be run in the host environment unless they should also service requests sent to jail IP addresses. Attempting to serve NFS from the host environment may also cause confusion, and cannot be easily reconfigured to use only specific IPs, as some NFS services are hosted directly from the kernel. Any third-party network software running in the host environment should also be checked and configured so that it does not bind all IP addresses, which would result in those services also appearing to be offered by the jail environments. .Pp Once these daemons have been disabled or fixed in the host environment, it is best to reboot so that all daemons are in a known state, to reduce the potential for confusion later (such as finding that when you send mail to a jail, and its sendmail is down, the mail is delivered to the host, etc.). .Ss "Configuring the Jail" Start any jail for the first time without configuring the network interface so that you can clean it up a little and set up accounts. As with any machine (virtual or not), you will need to set a root password, time zone, etc. Some of these steps apply only if you intend to run a full virtual server inside the jail; others apply both for constraining a particular application or for running a virtual server. .Pp Start a shell in the jail: .Bd -literal -offset indent jail -c path=/data/jail/testjail mount.devfs \\ host.hostname=testhostname ip4.addr=192.0.2.100 \\ command=/bin/sh .Ed .Pp Assuming no errors, you will end up with a shell prompt within the jail. You can now run .Pa /usr/sbin/bsdinstall and do the post-install configuration to set various configuration options, or perform these actions manually by editing .Pa /etc/rc.conf , etc. .Pp .Bl -bullet -offset indent -compact .It Configure .Pa /etc/resolv.conf so that name resolution within the jail will work correctly. .It Run .Xr newaliases 1 to quell .Xr sendmail 8 warnings. .It Set a root password, probably different from the real host system. .It Set the timezone. .It Add accounts for users in the jail environment. .It Install any packages the environment requires. .El .Pp You may also want to perform any package-specific configuration (web servers, SSH servers, etc), patch up .Pa /etc/syslog.conf so it logs as you would like, etc. If you are not using a virtual server, you may wish to modify .Xr syslogd 8 in the host environment to listen on the syslog socket in the jail environment; in this example, the syslog socket would be stored in .Pa /data/jail/testjail/var/run/log . .Pp Exit from the shell, and the jail will be shut down. .Ss "Starting the Jail" You are now ready to restart the jail and bring up the environment with all of its daemons and other programs. Create an entry for the jail in .Pa /etc/jail.conf : .Bd -literal -offset indent testjail { path = /tmp/jail/testjail; mount.devfs; host.hostname = testhostname; ip4.addr = 192.0.2.100; interface = ed0; exec.start = "/bin/sh /etc/rc"; exec.stop = "/bin/sh /etc/rc.shutdown"; } .Ed .Pp To start a virtual server environment, .Pa /etc/rc is run to launch various daemons and services, and .Pa /etc/rc.shutdown is run to shut them down when the jail is removed. If you are running a single application in the jail, substitute the command used to start the application for .Dq /bin/sh /etc/rc ; there may be some script available to cleanly shut down the application, or it may be sufficient to go without a stop command, and have .Nm send .Dv SIGTERM to the application. .Pp Start the jail by running: .Bd -literal -offset indent jail -c testjail .Ed .Pp A few warnings may be produced; however, it should all work properly. You should be able to see .Xr inetd 8 , .Xr syslogd 8 , and other processes running within the jail using .Xr ps 1 , with the .Ql J flag appearing beside jailed processes. To see an active list of jails, use .Xr jls 8 . If .Xr sshd 8 is enabled in the jail environment, you should be able to .Xr ssh 1 to the hostname or IP address of the jailed environment, and log in using the accounts you created previously. .Pp It is possible to have jails started at boot time. Please refer to the .Dq jail_* variables in .Xr rc.conf 5 for more information. .Ss "Managing the Jail" Normal machine shutdown commands, such as .Xr halt 8 , .Xr reboot 8 , and .Xr shutdown 8 , cannot be used successfully within the jail. To kill all processes from within a jail, you may use one of the following commands, depending on what you want to accomplish: .Bd -literal -offset indent kill -TERM -1 kill -KILL -1 .Ed .Pp This will send the .Dv SIGTERM or .Dv SIGKILL signals to all processes in the jail \(em be careful not to run this from the host environment! Once all of the jail's processes have died, unless the jail was created with the .Va persist parameter, the jail will be removed. Depending on the intended use of the jail, you may also want to run .Pa /etc/rc.shutdown from within the jail. .Pp To shut down the jail from the outside, simply remove it with .Nm .Ar -r , which will run any commands specified by .Va exec.stop , and then send .Dv SIGTERM and eventually .Dv SIGKILL to any remaining jailed processes. .Pp The .Pa /proc/ Ns Ar pid Ns Pa /status file contains, as its last field, the name of the jail in which the process runs, or .Dq Li - to indicate that the process is not running within a jail. The .Xr ps 1 command also shows a .Ql J flag for processes in a jail. .Pp You can also list/kill processes based on their jail ID. To show processes and their jail ID, use the following command: .Pp .Dl "ps ax -o pid,jid,args" .Pp To show and then kill processes in jail number 3 use the following commands: .Bd -literal -offset indent pgrep -lfj 3 pkill -j 3 .Ed or: .Pp .Dl "killall -j 3" .Ss "Jails and File Systems" It is not possible to .Xr mount 8 or .Xr umount 8 any file system inside a jail unless the file system is marked jail-friendly, the jail's .Va allow.mount parameter is set, and the jail's .Va enforce_statfs parameter is lower than 2. .Pp Multiple jails sharing the same file system can influence each other. For example, a user in one jail can fill the file system, leaving no space for processes in the other jail. Trying to use .Xr quota 1 to prevent this will not work either, as the file system quotas are not aware of jails but only look at the user and group IDs. This means the same user ID in two jails share a single file system quota. One would need to use one file system per jail to make this work. .Ss "Sysctl MIB Entries" The read-only entry .Va security.jail.jailed can be used to determine if a process is running inside a jail (value is one) or not (value is zero). .Pp The variable .Va security.jail.max_af_ips determines how may address per address family a jail may have. The default is 255. .Pp Some MIB variables have per-jail settings. Changes to these variables by a jailed process do not affect the host environment, only the jail environment. These variables are .Va kern.securelevel , .Va kern.hostname , .Va kern.domainname , .Va kern.hostid , and .Va kern.hostuuid . .Ss "Hierarchical Jails" By setting a jail's .Va children.max parameter, processes within a jail may be able to create jails of their own. These child jails are kept in a hierarchy, with jails only able to see and/or modify the jails they created (or those jails' children). Each jail has a read-only .Va parent parameter, containing the .Va jid of the jail that created it; a .Va jid of 0 indicates the jail is a child of the current jail (or is a top-level jail if the current process isn't jailed). .Pp Jailed processes are not allowed to confer greater permissions than they themselves are given, e.g., if a jail is created with .Va allow.nomount , it is not able to create a jail with .Va allow.mount set. Similarly, such restrictions as .Va ip4.addr and .Va securelevel may not be bypassed in child jails. .Pp A child jail may in turn create its own child jails if its own .Va children.max parameter is set (remember it is zero by default). These jails are visible to and can be modified by their parent and all ancestors. .Pp Jail names reflect this hierarchy, with a full name being an MIB-type string separated by dots. For example, if a base system process creates a jail .Dq foo , and a process under that jail creates another jail .Dq bar , then the second jail will be seen as .Dq foo.bar in the base system (though it is only seen as .Dq bar to any processes inside jail .Dq foo ) . Jids on the other hand exist in a single space, and each jail must have a unique jid. .Pp Like the names, a child jail's .Va path appears relative to its creator's own .Va path . This is by virtue of the child jail being created in the chrooted environment of the first jail. .Sh SEE ALSO .Xr killall 1 , .Xr lsvfs 1 , .Xr newaliases 1 , .Xr pgrep 1 , .Xr pkill 1 , .Xr ps 1 , .Xr quota 1 , .Xr jail_set 2 , +.Xr vmm 4 , .Xr devfs 5 , .Xr fdescfs 5 , .Xr jail.conf 5 , .Xr linprocfs 5 , .Xr linsysfs 5 , .Xr procfs 5 , .Xr rc.conf 5 , .Xr sysctl.conf 5 , .Xr chroot 8 , .Xr devfs 8 , .Xr halt 8 , .Xr ifconfig 8 , .Xr inetd 8 , .Xr jexec 8 , .Xr jls 8 , .Xr mount 8 , .Xr named 8 , .Xr reboot 8 , .Xr rpcbind 8 , .Xr sendmail 8 , .Xr shutdown 8 , .Xr sysctl 8 , .Xr syslogd 8 , .Xr umount 8 .Sh HISTORY The .Nm utility appeared in .Fx 4.0 . Hierarchical/extensible jails were introduced in .Fx 8.0 . The configuration file was introduced in .Fx 9.1 . .Sh AUTHORS .An -nosplit The jail feature was written by .An Poul-Henning Kamp for R&D Associates who contributed it to .Fx . .Pp .An Robert Watson wrote the extended documentation, found a few bugs, added a few new features, and cleaned up the userland jail environment. .Pp .An Bjoern A. Zeeb added multi-IP jail support for IPv4 and IPv6 based on a patch originally done by .An Pawel Jakub Dawidek for IPv4. .Pp .An James Gritton added the extensible jail parameters, hierarchical jails, and the configuration file. .Sh BUGS It might be a good idea to add an address alias flag such that daemons listening on all IPs .Pq Dv INADDR_ANY will not bind on that address, which would facilitate building a safe host environment such that host daemons do not impose on services offered from within jails. Currently, the simplest answer is to minimize services offered on the host, possibly limiting it to services offered from .Xr inetd 8 which is easily configurable. .Sh NOTES Great care should be taken when managing directories visible within the jail. For example, if a jailed process has its current working directory set to a directory that is moved out of the jail's chroot, then the process may gain access to the file space outside of the jail. It is recommended that directories always be copied, rather than moved, out of a jail. .Pp In addition, there are several ways in which an unprivileged user outside the jail can cooperate with a privileged user inside the jail and thereby obtain elevated privileges in the host environment. Most of these attacks can be mitigated by ensuring that the jail root is not accessible to unprivileged users in the host environment. Regardless, as a general rule, untrusted users with privileged access to a jail should not be given access to the host environment.