/* $OpenBSD: kern_unveil.c,v 1.54 2022/08/14 01:58:27 jsg Exp $ */ /* * Copyright (c) 2017-2019 Bob Beck * * Permission to use, copy, modify, and distribute this software for any * purpose with or without fee is hereby granted, provided that the above * copyright notice and this permission notice appear in all copies. * * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include struct unvname { char *un_name; size_t un_namesize; u_char un_flags; RBT_ENTRY(unvnmae) un_rbt; }; RBT_HEAD(unvname_rbt, unvname); struct unveil { struct vnode *uv_vp; ssize_t uv_cover; struct unvname_rbt uv_names; struct rwlock uv_lock; u_char uv_flags; }; /* #define DEBUG_UNVEIL */ #ifdef DEBUG_UNVEIL #define DPRINTF(x...) do { printf(x); } while (0) #else #define DPRINTF(x...) #endif #define UNVEIL_MAX_VNODES 128 #define UNVEIL_MAX_NAMES 128 static inline int unvname_compare(const struct unvname *n1, const struct unvname *n2) { if (n1->un_namesize == n2->un_namesize) return (memcmp(n1->un_name, n2->un_name, n1->un_namesize)); else return (n1->un_namesize - n2->un_namesize); } struct unvname * unvname_new(const char *name, size_t size, u_char flags) { struct unvname *ret = malloc(sizeof(struct unvname), M_PROC, M_WAITOK); ret->un_name = malloc(size, M_PROC, M_WAITOK); memcpy(ret->un_name, name, size); ret->un_namesize = size; ret->un_flags = flags; return ret; } void unvname_delete(struct unvname *name) { free(name->un_name, M_PROC, name->un_namesize); free(name, M_PROC, sizeof(struct unvname)); } RBT_PROTOTYPE(unvname_rbt, unvname, un_rbt, unvname_compare); RBT_GENERATE(unvname_rbt, unvname, un_rbt, unvname_compare); int unveil_delete_names(struct unveil *uv) { struct unvname *unvn, *next; int ret = 0; rw_enter_write(&uv->uv_lock); RBT_FOREACH_SAFE(unvn, unvname_rbt, &uv->uv_names, next) { RBT_REMOVE(unvname_rbt, &uv->uv_names, unvn); unvname_delete(unvn); ret++; } rw_exit_write(&uv->uv_lock); DPRINTF("deleted %d names\n", ret); return ret; } int unveil_add_name_unlocked(struct unveil *uv, char *name, u_char flags) { struct unvname *unvn; unvn = unvname_new(name, strlen(name) + 1, flags); if (RBT_INSERT(unvname_rbt, &uv->uv_names, unvn) != NULL) { /* Name already present. */ unvname_delete(unvn); return 0; } DPRINTF("added name %s underneath vnode %p\n", name, uv->uv_vp); return 1; } int unveil_add_name(struct unveil *uv, char *name, u_char flags) { int ret; rw_enter_write(&uv->uv_lock); ret = unveil_add_name_unlocked(uv, name, flags); rw_exit_write(&uv->uv_lock); return ret; } struct unvname * unveil_namelookup(struct unveil *uv, char *name) { struct unvname n, *ret = NULL; rw_enter_read(&uv->uv_lock); DPRINTF("%s: looking up name %s (%p) in vnode %p\n", __func__, name, name, uv->uv_vp); KASSERT(uv->uv_vp != NULL); n.un_name = name; n.un_namesize = strlen(name) + 1; ret = RBT_FIND(unvname_rbt, &uv->uv_names, &n); rw_exit_read(&uv->uv_lock); DPRINTF("%s: %s name %s in vnode %p\n", __func__, (ret == NULL) ? "no match for" : "matched", name, uv->uv_vp); return ret; } void unveil_destroy(struct process *ps) { size_t i; for (i = 0; ps->ps_uvpaths != NULL && i < ps->ps_uvvcount; i++) { struct unveil *uv = ps->ps_uvpaths + i; struct vnode *vp = uv->uv_vp; /* skip any vnodes zapped by unveil_removevnode */ if (vp != NULL) { vp->v_uvcount--; DPRINTF("unveil: %s(%d): removing vnode %p uvcount %d " "in position %ld\n", ps->ps_comm, ps->ps_pid, vp, vp->v_uvcount, i); vrele(vp); } ps->ps_uvncount -= unveil_delete_names(uv); uv->uv_vp = NULL; uv->uv_flags = 0; } KASSERT(ps->ps_uvncount == 0); free(ps->ps_uvpaths, M_PROC, UNVEIL_MAX_VNODES * sizeof(struct unveil)); ps->ps_uvvcount = 0; ps->ps_uvpaths = NULL; } void unveil_copy(struct process *parent, struct process *child) { size_t i; child->ps_uvdone = parent->ps_uvdone; if (parent->ps_uvvcount == 0) return; child->ps_uvpaths = mallocarray(UNVEIL_MAX_VNODES, sizeof(struct unveil), M_PROC, M_WAITOK|M_ZERO); child->ps_uvncount = 0; for (i = 0; parent->ps_uvpaths != NULL && i < parent->ps_uvvcount; i++) { struct unveil *from = parent->ps_uvpaths + i; struct unveil *to = child->ps_uvpaths + i; struct unvname *unvn, *next; to->uv_vp = from->uv_vp; if (to->uv_vp != NULL) { vref(to->uv_vp); to->uv_vp->v_uvcount++; } rw_init(&to->uv_lock, "unveil"); RBT_INIT(unvname_rbt, &to->uv_names); rw_enter_read(&from->uv_lock); RBT_FOREACH_SAFE(unvn, unvname_rbt, &from->uv_names, next) { if (unveil_add_name_unlocked(&child->ps_uvpaths[i], unvn->un_name, unvn->un_flags)) child->ps_uvncount++; } rw_exit_read(&from->uv_lock); to->uv_flags = from->uv_flags; to->uv_cover = from->uv_cover; } child->ps_uvvcount = parent->ps_uvvcount; } /* * Walk up from vnode dp, until we find a matching unveil, or the root vnode * returns -1 if no unveil to be found above dp or if dp is the root vnode. */ ssize_t unveil_find_cover(struct vnode *dp, struct proc *p) { struct vnode *vp = NULL, *parent = NULL, *root; ssize_t ret = -1; int error; /* use the correct root to stop at, chrooted or not.. */ root = p->p_fd->fd_rdir ? p->p_fd->fd_rdir : rootvnode; vp = dp; while (vp != root) { struct componentname cn = { .cn_nameiop = LOOKUP, .cn_flags = ISLASTCN | ISDOTDOT | RDONLY, .cn_proc = p, .cn_cred = p->p_ucred, .cn_pnbuf = NULL, .cn_nameptr = "..", .cn_namelen = 2, .cn_consume = 0 }; /* * If we are at the root of a filesystem, and we are * still mounted somewhere, take the .. in the above * filesystem. */ if (vp != root && (vp->v_flag & VROOT)) { if (vp->v_mount == NULL) return -1; vp = vp->v_mount->mnt_vnodecovered ? vp->v_mount->mnt_vnodecovered : vp; } if (vget(vp, LK_EXCLUSIVE|LK_RETRY) != 0) return -1; /* Get parent vnode of vp using lookup of '..' */ /* This returns with vp unlocked but ref'ed*/ error = VOP_LOOKUP(vp, &parent, &cn); if (error) { if (!(cn.cn_flags & PDIRUNLOCK)) vput(vp); else { /* * This corner case should not happen because * we have not set LOCKPARENT in the flags */ DPRINTF("vnode %p PDIRUNLOCK on error\n", vp); vrele(vp); } break; } vrele(vp); (void) unveil_lookup(parent, p->p_p, &ret); vput(parent); if (ret >= 0) break; if (vp == parent) { ret = -1; break; } vp = parent; parent = NULL; } return ret; } struct unveil * unveil_lookup(struct vnode *vp, struct process *pr, ssize_t *position) { struct unveil *uv = pr->ps_uvpaths; ssize_t i; if (position != NULL) *position = -1; if (vp->v_uvcount == 0) return NULL; for (i = 0; i < pr->ps_uvvcount; i++) { if (vp == uv[i].uv_vp) { KASSERT(uv[i].uv_vp->v_uvcount > 0); KASSERT(uv[i].uv_vp->v_usecount > 0); if (position != NULL) *position = i; return &uv[i]; } } return NULL; } int unveil_parsepermissions(const char *permissions, u_char *perms) { size_t i = 0; char c; *perms = UNVEIL_USERSET; while ((c = permissions[i++]) != '\0') { switch (c) { case 'r': *perms |= UNVEIL_READ; break; case 'w': *perms |= UNVEIL_WRITE; break; case 'x': *perms |= UNVEIL_EXEC; break; case 'c': *perms |= UNVEIL_CREATE; break; default: return -1; } } return 0; } int unveil_setflags(u_char *flags, u_char nflags) { #if 0 if (((~(*flags)) & nflags) != 0) { DPRINTF("Flags escalation %llX -> %llX\n", *flags, nflags); return 1; } #endif *flags = nflags; return 1; } struct unveil * unveil_add_vnode(struct proc *p, struct vnode *vp) { struct process *pr = p->p_p; struct unveil *uv = NULL; ssize_t i; KASSERT(pr->ps_uvvcount < UNVEIL_MAX_VNODES); uv = &pr->ps_uvpaths[pr->ps_uvvcount++]; rw_init(&uv->uv_lock, "unveil"); RBT_INIT(unvname_rbt, &uv->uv_names); uv->uv_vp = vp; uv->uv_flags = 0; /* find out what we are covered by */ uv->uv_cover = unveil_find_cover(vp, p); /* * Find anyone covered by what we are covered by * and re-check what covers them (we could have * interposed a cover) */ for (i = 0; i < pr->ps_uvvcount - 1; i++) { if (pr->ps_uvpaths[i].uv_cover == uv->uv_cover) pr->ps_uvpaths[i].uv_cover = unveil_find_cover(pr->ps_uvpaths[i].uv_vp, p); } return (uv); } int unveil_add(struct proc *p, struct nameidata *ndp, const char *permissions) { struct process *pr = p->p_p; struct vnode *vp; struct unveil *uv; int directory_add; int ret = EINVAL; u_char flags; KASSERT(ISSET(ndp->ni_cnd.cn_flags, HASBUF)); /* must have SAVENAME */ if (unveil_parsepermissions(permissions, &flags) == -1) goto done; if (pr->ps_uvpaths == NULL) { pr->ps_uvpaths = mallocarray(UNVEIL_MAX_VNODES, sizeof(struct unveil), M_PROC, M_WAITOK|M_ZERO); } if (pr->ps_uvvcount >= UNVEIL_MAX_VNODES || pr->ps_uvncount >= UNVEIL_MAX_NAMES) { ret = E2BIG; goto done; } /* Are we a directory? or something else */ directory_add = ndp->ni_vp != NULL && ndp->ni_vp->v_type == VDIR; if (directory_add) vp = ndp->ni_vp; else vp = ndp->ni_dvp; KASSERT(vp->v_type == VDIR); vref(vp); vp->v_uvcount++; if ((uv = unveil_lookup(vp, pr, NULL)) != NULL) { /* * We already have unveiled this directory * vnode */ vp->v_uvcount--; vrele(vp); /* * If we are adding a directory which was already * unveiled containing only specific terminals, * unrestrict it. */ if (directory_add) { DPRINTF("unveil: %s(%d): updating directory vnode %p" " to unrestricted uvcount %d\n", pr->ps_comm, pr->ps_pid, vp, vp->v_uvcount); if (!unveil_setflags(&uv->uv_flags, flags)) ret = EPERM; else ret = 0; goto done; } /* * If we are adding a terminal that is already unveiled, just * replace the flags and we are done */ if (!directory_add) { struct unvname *tname; if ((tname = unveil_namelookup(uv, ndp->ni_cnd.cn_nameptr)) != NULL) { DPRINTF("unveil: %s(%d): changing flags for %s" "in vnode %p, uvcount %d\n", pr->ps_comm, pr->ps_pid, tname->un_name, vp, vp->v_uvcount); if (!unveil_setflags(&tname->un_flags, flags)) ret = EPERM; else ret = 0; goto done; } } } else { /* * New unveil involving this directory vnode. */ uv = unveil_add_vnode(p, vp); } /* * At this stage with have a unveil in uv with a vnode for a * directory. If the component we are adding is a directory, * we are done. Otherwise, we add the component name the name * list in uv. */ if (directory_add) { uv->uv_flags = flags; ret = 0; DPRINTF("unveil: %s(%d): added unrestricted directory vnode %p" ", uvcount %d\n", pr->ps_comm, pr->ps_pid, vp, vp->v_uvcount); goto done; } if (unveil_add_name(uv, ndp->ni_cnd.cn_nameptr, flags)) pr->ps_uvncount++; ret = 0; DPRINTF("unveil: %s(%d): added name %s beneath %s vnode %p," " uvcount %d\n", pr->ps_comm, pr->ps_pid, ndp->ni_cnd.cn_nameptr, uv->uv_flags ? "unrestricted" : "restricted", vp, vp->v_uvcount); done: return ret; } /* * XXX this will probably change. * XXX collapse down later once debug surely unneeded */ int unveil_flagmatch(struct nameidata *ni, u_char flags) { if (flags == 0) { DPRINTF("All operations forbidden for 0 flags\n"); return 0; } if (ni->ni_unveil & UNVEIL_READ) { if ((flags & UNVEIL_READ) == 0) { DPRINTF("unveil lacks UNVEIL_READ\n"); return 0; } } if (ni->ni_unveil & UNVEIL_WRITE) { if ((flags & UNVEIL_WRITE) == 0) { DPRINTF("unveil lacks UNVEIL_WRITE\n"); return 0; } } if (ni->ni_unveil & UNVEIL_EXEC) { if ((flags & UNVEIL_EXEC) == 0) { DPRINTF("unveil lacks UNVEIL_EXEC\n"); return 0; } } if (ni->ni_unveil & UNVEIL_CREATE) { if ((flags & UNVEIL_CREATE) == 0) { DPRINTF("unveil lacks UNVEIL_CREATE\n"); return 0; } } return 1; } /* * When traversing up towards the root figure out the proper unveil for * the parent directory. */ struct unveil * unveil_covered(struct unveil *uv, struct vnode *dvp, struct proc *p) { if (uv && uv->uv_vp == dvp) { /* if at the root, chrooted or not, return the current uv */ if (dvp == (p->p_fd->fd_rdir ? p->p_fd->fd_rdir : rootvnode)) return uv; if (uv->uv_cover >=0) { KASSERT(uv->uv_cover < p->p_p->ps_uvvcount); return &p->p_p->ps_uvpaths[uv->uv_cover]; } return NULL; } return uv; } /* * Start a relative path lookup. Ensure we find whatever unveil covered * where we start from, either by having a saved current working directory * unveil, or by walking up and finding a cover the hard way if we are * doing a non AT_FDCWD relative lookup. Caller passes a NULL dp * if we are using AT_FDCWD. */ void unveil_start_relative(struct proc *p, struct nameidata *ni, struct vnode *dp) { struct process *pr = p->p_p; struct unveil *uv = NULL; ssize_t uvi; if (pr->ps_uvpaths == NULL) return; uv = unveil_lookup(dp, pr, NULL); if (uv == NULL) { uvi = unveil_find_cover(dp, p); if (uvi >= 0) { KASSERT(uvi < pr->ps_uvvcount); uv = &pr->ps_uvpaths[uvi]; } } /* * Store this match for later use. Flags are checked at the end. */ if (uv) { DPRINTF("unveil: %s(%d): relative unveil at %p matches", pr->ps_comm, pr->ps_pid, uv); ni->ni_unveil_match = uv; } } /* * unveil checking - for component directories in a namei lookup. */ void unveil_check_component(struct proc *p, struct nameidata *ni, struct vnode *dp) { struct process *pr = p->p_p; struct unveil *uv = NULL; if (ni->ni_pledge == PLEDGE_UNVEIL || pr->ps_uvpaths == NULL) return; if (ni->ni_cnd.cn_flags & BYPASSUNVEIL) return; if (ni->ni_cnd.cn_flags & ISDOTDOT) { /* * adjust unveil match as necessary */ uv = unveil_covered(ni->ni_unveil_match, dp, p); /* clear the match when we DOTDOT above it */ if (ni->ni_unveil_match && ni->ni_unveil_match->uv_vp == dp) ni->ni_unveil_match = NULL; } else uv = unveil_lookup(dp, pr, NULL); if (uv != NULL) { /* update match */ ni->ni_unveil_match = uv; DPRINTF("unveil: %s(%d): component directory match for " "vnode %p\n", pr->ps_comm, pr->ps_pid, dp); } } /* * unveil checking - only done after namei lookup has succeeded on * the last component of a namei lookup. */ int unveil_check_final(struct proc *p, struct nameidata *ni) { struct process *pr = p->p_p; struct unveil *uv = NULL, *nuv; struct unvname *tname = NULL; if (ni->ni_pledge == PLEDGE_UNVEIL || pr->ps_uvpaths == NULL) return (0); if (ni->ni_cnd.cn_flags & BYPASSUNVEIL) { DPRINTF("unveil: %s(%d): BYPASSUNVEIL.\n", pr->ps_comm, pr->ps_pid); return (0); } if (ni->ni_vp != NULL && ni->ni_vp->v_type == VDIR) { /* We are matching a directory terminal component */ uv = unveil_lookup(ni->ni_vp, pr, NULL); if (uv == NULL || (uv->uv_flags & UNVEIL_USERSET) == 0) { DPRINTF("unveil: %s(%d) no match for vnode %p\n", pr->ps_comm, pr->ps_pid, ni->ni_vp); if (uv != NULL) ni->ni_unveil_match = uv; goto done; } if (!unveil_flagmatch(ni, uv->uv_flags)) { DPRINTF("unveil: %s(%d) flag mismatch for directory" " vnode %p\n", pr->ps_comm, pr->ps_pid, ni->ni_vp); pr->ps_acflag |= AUNVEIL; if (uv->uv_flags & UNVEIL_MASK) return EACCES; else return ENOENT; } /* directory and flags match, success */ DPRINTF("unveil: %s(%d): matched directory \"%s\" at vnode %p\n", pr->ps_comm, pr->ps_pid, ni->ni_cnd.cn_nameptr, uv->uv_vp); return (0); } /* Otherwise, we are matching a non-terminal component */ uv = unveil_lookup(ni->ni_dvp, pr, NULL); if (uv == NULL) { DPRINTF("unveil: %s(%d) no match for directory vnode %p\n", pr->ps_comm, pr->ps_pid, ni->ni_dvp); goto done; } if ((tname = unveil_namelookup(uv, ni->ni_cnd.cn_nameptr)) == NULL) { DPRINTF("unveil: %s(%d) no match for terminal '%s' in " "directory vnode %p\n", pr->ps_comm, pr->ps_pid, ni->ni_cnd.cn_nameptr, ni->ni_dvp); /* no specific name, so check unveil directory flags */ if (!unveil_flagmatch(ni, uv->uv_flags)) { DPRINTF("unveil: %s(%d) terminal " "'%s' flags mismatch in directory " "vnode %p\n", pr->ps_comm, pr->ps_pid, ni->ni_cnd.cn_nameptr, ni->ni_dvp); /* * If dir has user set restrictions fail with * EACCES or ENOENT. Otherwise, use any covering * match that we found above this dir. */ if (uv->uv_flags & UNVEIL_USERSET) { pr->ps_acflag |= AUNVEIL; if (uv->uv_flags & UNVEIL_MASK) return EACCES; else return ENOENT; } /* start backtrack from this node */ ni->ni_unveil_match = uv; goto done; } /* directory flags match, success */ DPRINTF("unveil: %s(%d): matched \"%s\" underneath vnode %p\n", pr->ps_comm, pr->ps_pid, ni->ni_cnd.cn_nameptr, uv->uv_vp); return (0); } if (!unveil_flagmatch(ni, tname->un_flags)) { /* do flags match for matched name */ DPRINTF("unveil: %s(%d) flag mismatch for terminal '%s'\n", pr->ps_comm, pr->ps_pid, tname->un_name); pr->ps_acflag |= AUNVEIL; return EACCES; } /* name and flags match. success */ DPRINTF("unveil: %s(%d) matched terminal '%s'\n", pr->ps_comm, pr->ps_pid, tname->un_name); return (0); done: /* * last component did not match, check previous matches if * access is allowed or not. */ for (uv = ni->ni_unveil_match; uv != NULL; uv = nuv) { if (unveil_flagmatch(ni, uv->uv_flags)) { DPRINTF("unveil: %s(%d): matched \"%s\" underneath/at " "vnode %p\n", pr->ps_comm, pr->ps_pid, ni->ni_cnd.cn_nameptr, uv->uv_vp); return (0); } /* if node has any flags set then this is an access violation */ if (uv->uv_flags & UNVEIL_USERSET) { DPRINTF("unveil: %s(%d) flag mismatch for vnode %p\n", pr->ps_comm, pr->ps_pid, uv->uv_vp); pr->ps_acflag |= AUNVEIL; if (uv->uv_flags & UNVEIL_MASK) return EACCES; else return ENOENT; } DPRINTF("unveil: %s(%d) check cover for vnode %p, uv_cover %zd\n", pr->ps_comm, pr->ps_pid, uv->uv_vp, uv->uv_cover); nuv = unveil_covered(uv, uv->uv_vp, p); if (nuv == uv) break; } pr->ps_acflag |= AUNVEIL; return ENOENT; } /* * Scan all active processes to see if any of them have a unveil * to this vnode. If so, NULL the vnode in their unveil list, * vrele, drop the reference, and mark their unveil list * as needing to have the hole shrunk the next time the process * uses it for lookup. */ void unveil_removevnode(struct vnode *vp) { struct process *pr; if (vp->v_uvcount == 0) return; DPRINTF("%s: found vnode %p with count %d\n", __func__, vp, vp->v_uvcount); vref(vp); /* make sure it is held till we are done */ LIST_FOREACH(pr, &allprocess, ps_list) { struct unveil * uv; if ((uv = unveil_lookup(vp, pr, NULL)) != NULL && uv->uv_vp != NULL) { uv->uv_vp = NULL; uv->uv_flags = 0; DPRINTF("%s: vnode %p now count %d\n", __func__, vp, vp->v_uvcount); if (vp->v_uvcount > 0) { vrele(vp); vp->v_uvcount--; } else panic("vp %p, v_uvcount of %d should be 0", vp, vp->v_uvcount); } } KASSERT(vp->v_uvcount == 0); vrele(vp); /* release our ref */ }