diff options
author | Ted Unangst <tedu@cvs.openbsd.org> | 2003-05-12 20:58:41 +0000 |
---|---|---|
committer | Ted Unangst <tedu@cvs.openbsd.org> | 2003-05-12 20:58:41 +0000 |
commit | a4037f6b4fd1ba42f6428bea41cfc9ff22c44528 (patch) | |
tree | 4013cabcc8e4b31ead3f6880165d3cb4684e028b | |
parent | 37667141ad0568061d95baa27233e1915c20c28a (diff) |
new files to support layered file systems. to be used in a bit.
mostly from work by Bill Studenmund - wrstuden at netbsd
-rw-r--r-- | sys/miscfs/genfs/layer.h | 167 | ||||
-rw-r--r-- | sys/miscfs/genfs/layer_extern.h | 122 | ||||
-rw-r--r-- | sys/miscfs/genfs/layer_subr.c | 401 | ||||
-rw-r--r-- | sys/miscfs/genfs/layer_vfsops.c | 285 | ||||
-rw-r--r-- | sys/miscfs/genfs/layer_vnops.c | 950 |
5 files changed, 1925 insertions, 0 deletions
diff --git a/sys/miscfs/genfs/layer.h b/sys/miscfs/genfs/layer.h new file mode 100644 index 00000000000..c2e4451c16e --- /dev/null +++ b/sys/miscfs/genfs/layer.h @@ -0,0 +1,167 @@ +/* $OpenBSD: layer.h,v 1.1 2003/05/12 20:58:40 tedu Exp $ */ +/* $NetBSD: layer.h,v 1.4 2001/06/07 13:32:46 wiz Exp $ */ + +/* + * Copyright (c) 1999 National Aeronautics & Space Administration + * All rights reserved. + * + * This software was written by William Studenmund of the + * Numerical Aerospace Simulation Facility, NASA Ames Research Center. + * + * 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. + * 3. Neither the name of the National Aeronautics & Space Administration + * nor the names of its contributors may be used to endorse or promote + * products derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY THE NATIONAL AERONAUTICS & SPACE ADMINISTRATION + * ``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 ADMINISTRATION OR CONTRIB- + * UTORS 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. + */ + +/* + * Copyright (c) 1992, 1993 + * The Regents of the University of California. All rights reserved. + * + * This code is derived from software donated to Berkeley by + * Jan-Simon Pendry. + * + * 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. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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. + * + * from: Id: lofs.h,v 1.8 1992/05/30 10:05:43 jsp Exp + * @(#)null.h 8.2 (Berkeley) 1/21/94 + */ + +#ifndef _MISCFS_GENFS_LAYER_H_ +#define _MISCFS_GENFS_LAYER_H_ + +struct layer_args { + char *target; /* Target of loopback */ + struct export_args export; /* network export info */ +}; + +#ifdef _KERNEL + +struct layer_node; + +LIST_HEAD(layer_node_hashhead, layer_node); + +struct layer_mount { + struct mount *layerm_vfs; + struct vnode *layerm_rootvp; /* Ref to root layer_node */ + struct netexport layerm_export; /* export info */ + u_int layerm_flags; /* mount point layer flags */ + u_int layerm_size; /* size of fs's struct node */ + enum vtype layerm_tag; /* vtag of our vnodes */ + int /* bypass routine for this mount */ + (*layerm_bypass)(void *); + int (*layerm_alloc) /* alloc a new layer node */ + (struct mount *, struct vnode *, + struct vnode **); + int /* ops for our nodes */ + (**layerm_vnodeop_p)(void *); + struct layer_node_hashhead /* head of hash list for layer_nodes */ + *layerm_node_hashtbl; + u_long layerm_node_hash; /* hash mask for hash chain */ + struct simplelock layerm_hashlock; /* interlock for hash chain. */ +}; + +#define LAYERFS_MFLAGS 0x00000fff /* reserved layer mount flags */ +#define LAYERFS_MBYPASSDEBUG 0x00000001 + +/* + * A cache of vnode references + */ +struct layer_node { + LIST_ENTRY(layer_node) layer_hash; /* Hash list */ + struct vnode *layer_lowervp; /* VREFed once */ + struct vnode *layer_vnode; /* Back pointer */ + unsigned int layer_flags; /* locking, etc. */ +}; + +#define LAYERFS_RESFLAGS 0x00000fff /* flags reserved for layerfs */ + +/* + * The following macros handle upperfs-specific locking. They are needed + * when the lowerfs does not export a struct lock for locking use by the + * upper layers. These macros are inteded for adjusting the upperfs + * struct lock to reflect changes in the underlying vnode's lock state. + */ +#define LAYERFS_UPPERLOCK(v, f, r) do { \ + if ((v)->v_vnlock == NULL) \ + r = lockmgr(&(v)->v_lock, (f), &(v)->v_interlock, curproc); \ + else \ + r = 0; \ + } while (0) + +#define LAYERFS_UPPERUNLOCK(v, f, r) do { \ + if ((v)->v_vnlock == NULL) \ + r = lockmgr(&(v)->v_lock, (f) | LK_RELEASE, &(v)->v_interlock, curproc); \ + else \ + r = 0; \ + } while (0) + +#define LAYERFS_UPPERISLOCKED(v, r) do { \ + if ((v)->v_vnlock == NULL) \ + r = lockstatus(&(v)->v_lock); \ + else \ + r = -1; \ + } while (0) + +#define LAYERFS_DO_BYPASS(vp, ap) \ + (*MOUNTTOLAYERMOUNT((vp)->v_mount)->layerm_bypass)((ap)) + +extern struct vnode *layer_checkvp(struct vnode *vp, char *fil, int lno); + +#define MOUNTTOLAYERMOUNT(mp) ((struct layer_mount *)((mp)->mnt_data)) +#define VTOLAYER(vp) ((struct layer_node *)(vp)->v_data) +#define LAYERTOV(xp) ((xp)->layer_vnode) +#ifdef LAYERFS_DIAGNOSTIC +#define LAYERVPTOLOWERVP(vp) layer_checkvp((vp), __FILE__, __LINE__) +#else +#define LAYERVPTOLOWERVP(vp) (VTOLAYER(vp)->layer_lowervp) +#endif + +#endif /* _KERNEL */ +#endif /* _MISCFS_GENFS_LAYER_H_ */ diff --git a/sys/miscfs/genfs/layer_extern.h b/sys/miscfs/genfs/layer_extern.h new file mode 100644 index 00000000000..892b637f0a7 --- /dev/null +++ b/sys/miscfs/genfs/layer_extern.h @@ -0,0 +1,122 @@ +/* $OpenBSD: layer_extern.h,v 1.1 2003/05/12 20:58:40 tedu Exp $ */ +/* $NetBSD: layer_extern.h,v 1.5 2001/12/06 04:29:23 chs Exp $ */ + +/* + * Copyright (c) 1999 National Aeronautics & Space Administration + * All rights reserved. + * + * This software was written by William Studenmund of the + * Numerical Aerospace Simulation Facility, NASA Ames Research Center. + * + * 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. + * 3. Neither the name of the National Aeronautics & Space Administration + * nor the names of its contributors may be used to endorse or promote + * products derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY THE NATIONAL AERONAUTICS & SPACE ADMINISTRATION + * ``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 ADMINISTRATION OR CONTRIB- + * UTORS 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. + */ +/* + * Copyright (c) 1992, 1993, 1995 + * The Regents of the University of California. All rights reserved. + * + * This code is derived from software donated to Berkeley by + * Jan-Simon Pendry. + * + * 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. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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. + * + */ + +/* + * Routines defined by layerfs + */ + +/* misc routines in layer_subr.c */ +int layerfs_init(struct vfsconf *); +void layerfs_done(void); +int layer_node_alloc(struct mount *, struct vnode *, struct vnode **); +int layer_node_create(struct mount *, struct vnode *, struct vnode **); +struct vnode * + layer_node_find(struct mount *, struct vnode *); +#define LOG2_SIZEVNODE 7 /* log2(sizeof struct vnode) */ +#define LAYER_NHASH(lmp, vp) \ + (&((lmp)->layerm_node_hashtbl[(((u_long)vp)>>LOG2_SIZEVNODE) & \ + (lmp)->layerm_node_hash])) + +/* vfs routines */ +int layerfs_start(struct mount *, int, struct proc *); +int layerfs_root(struct mount *, struct vnode **); +int layerfs_quotactl(struct mount *, int, uid_t, caddr_t, + struct proc *); +int layerfs_statfs(struct mount *, struct statfs *, struct proc *); +int layerfs_sync(struct mount *, int, struct ucred *, struct proc *); +int layerfs_vget(struct mount *, ino_t, struct vnode **); +int layerfs_fhtovp(struct mount *, struct fid *, struct vnode **); +int layerfs_checkexp(struct mount *, struct mbuf *, int *, + struct ucred **); +int layerfs_vptofh(struct vnode *, struct fid *); +int layerfs_sysctl(int *, u_int, void *, size_t *, void *, size_t, + struct proc *); + +/* VOP routines */ +int layer_bypass(void *); +int layer_getattr(void *); +int layer_inactive(void *); +int layer_reclaim(void *); +int layer_print(void *); +int layer_strategy(void *); +int layer_bwrite(void *); +int layer_bmap(void *); +int layer_lock(void *); +int layer_unlock(void *); +int layer_islocked(void *); +int layer_fsync(void *); +int layer_lookup(void *); +int layer_setattr(void *); +int layer_access(void *); +int layer_open(void *); +int layer_getpages(void *); +int layer_putpages(void *); diff --git a/sys/miscfs/genfs/layer_subr.c b/sys/miscfs/genfs/layer_subr.c new file mode 100644 index 00000000000..5b5ed9eaaac --- /dev/null +++ b/sys/miscfs/genfs/layer_subr.c @@ -0,0 +1,401 @@ +/* $OpenBSD: layer_subr.c,v 1.1 2003/05/12 20:58:40 tedu Exp $ */ +/* $NetBSD: layer_subr.c,v 1.11 2002/02/20 06:16:22 enami Exp $ */ + +/* + * Copyright (c) 1999 National Aeronautics & Space Administration + * All rights reserved. + * + * This software was written by William Studenmund of the + * Numerical Aerospace Simulation Facility, NASA Ames Research Center. + * + * 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. + * 3. Neither the name of the National Aeronautics & Space Administration + * nor the names of its contributors may be used to endorse or promote + * products derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY THE NATIONAL AERONAUTICS & SPACE ADMINISTRATION + * ``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 ADMINISTRATION OR CONTRIB- + * UTORS 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. + */ +/* + * Copyright (c) 1992, 1993 + * The Regents of the University of California. All rights reserved. + * + * This code is derived from software donated to Berkeley by + * Jan-Simon Pendry. + * + * 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. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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. + * + * from: Id: lofs_subr.c,v 1.11 1992/05/30 10:05:43 jsp Exp + * @(#)null_subr.c 8.7 (Berkeley) 5/14/95 + */ + + +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/proc.h> +#include <sys/time.h> +#include <sys/vnode.h> +#include <sys/mount.h> +#include <sys/namei.h> +#include <sys/malloc.h> +#include <miscfs/specfs/specdev.h> +#include <miscfs/genfs/layer.h> +#include <miscfs/genfs/layer_extern.h> + +#define NLAYERNODECACHE 16 + +/* + * layer cache: + * Each cache entry holds a reference to the lower vnode + * along with a pointer to the alias vnode. When an + * entry is added the lower vnode is VREF'd. When the + * alias is removed the lower vnode is vrele'd. + */ + +/* + * Initialise cache headers + */ +int +layerfs_init(struct vfsconf *v) +{ +#ifdef LAYERFS_DIAGNOSTIC + printf("layerfs_init\n"); /* printed during system boot */ +#endif + return (0); +} + +/* + * Free global resources of layerfs. + */ +void +layerfs_done() +{ +#ifdef LAYERFS_DIAGNOSTIC + printf("layerfs_done\n"); /* printed on layerfs detach */ +#endif +} + +/* + * Return a locked, VREF'ed alias for lower vnode if already exists, else 0. + */ +struct vnode * +layer_node_find(mp, lowervp) + struct mount *mp; + struct vnode *lowervp; +{ + struct layer_mount *lmp = MOUNTTOLAYERMOUNT(mp); + struct layer_node_hashhead *hd; + struct layer_node *a; + struct vnode *vp; + + /* + * Find hash base, and then search the (two-way) linked + * list looking for a layer_node structure which is referencing + * the lower vnode. If found, the increment the layer_node + * reference count (but NOT the lower vnode's VREF counter) + * and return the vnode locked. + */ + hd = LAYER_NHASH(lmp, lowervp); +loop: + simple_lock(&lmp->layerm_hashlock); + for (a = hd->lh_first; a != 0; a = a->layer_hash.le_next) { + if (a->layer_lowervp == lowervp && LAYERTOV(a)->v_mount == mp) { + vp = LAYERTOV(a); + simple_unlock(&lmp->layerm_hashlock); + /* + * We must be careful here as the fact the lower + * vnode is locked will imply vp is locked unless + * someone has decided to start vclean'ing either + * vp or lowervp. + * + * So we try for an exclusive, recursive lock + * on the upper vnode. If it fails, vcleaning + * is in progress (so when we try again, we'll + * fail). If it succeeds, we now have double + * locked the bottom node. So we do an explicit + * VOP_UNLOCK on it to keep the counts right. Note + * that we will end up with the upper node and + * the lower node locked once. + */ + if (vget(vp, LK_EXCLUSIVE | LK_CANRECURSE, curproc)) { + printf ("layer_node_find: vget failed.\n"); + goto loop; + } + VOP_UNLOCK(lowervp, 0, curproc); + return (vp); + } + } + + simple_unlock(&lmp->layerm_hashlock); + return NULL; +} + + +/* + * Make a new layer_node node. + * Vp is the alias vnode, lowervp is the lower vnode. + * Maintain a reference to lowervp. + */ +int +layer_node_alloc(mp, lowervp, vpp) + struct mount *mp; + struct vnode *lowervp; + struct vnode **vpp; +{ + struct layer_mount *lmp = MOUNTTOLAYERMOUNT(mp); + struct layer_node_hashhead *hd; + struct layer_node *xp; + struct vnode *vp, *nvp; + int error; + extern int (**dead_vnodeop_p)(void *); + + if ((error = getnewvnode(lmp->layerm_tag, mp, lmp->layerm_vnodeop_p, + &vp)) != 0) + return (error); + vp->v_type = lowervp->v_type; + vp->v_flag |= VLAYER; + + xp = malloc(lmp->layerm_size, M_TEMP, M_WAITOK); + if (vp->v_type == VBLK || vp->v_type == VCHR) { + MALLOC(vp->v_specinfo, struct specinfo *, + sizeof(struct specinfo), M_VNODE, M_WAITOK); + vp->v_hashchain = NULL; + vp->v_rdev = lowervp->v_rdev; + } + + vp->v_data = xp; + xp->layer_vnode = vp; + xp->layer_lowervp = lowervp; + xp->layer_flags = 0; + /* + * Before we insert our new node onto the hash chains, + * check to see if someone else has beaten us to it. + * (We could have slept in MALLOC.) + */ + if ((nvp = layer_node_find(mp, lowervp)) != NULL) { + *vpp = nvp; + + /* free the substructures we've allocated. */ + FREE(xp, M_TEMP); + if (vp->v_type == VBLK || vp->v_type == VCHR) + FREE(vp->v_specinfo, M_VNODE); + + vp->v_type = VBAD; /* node is discarded */ + vp->v_op = dead_vnodeop_p; /* so ops will still work */ + vrele(vp); /* get rid of it. */ + return (0); + } + + simple_lock(&lmp->layerm_hashlock); + + /* + * Now lock the new node. We rely on the fact that we were passed + * a locked vnode. If the lower node is exporting a struct lock + * (v_vnlock != NULL) then we just set the upper v_vnlock to the + * lower one, and both are now locked. If the lower node is exporting + * NULL, then we copy that up and manually lock the upper node. + * + * LAYERFS_UPPERLOCK already has the test, so we use it after copying + * up the v_vnlock from below. + */ + + vp->v_vnlock = lowervp->v_vnlock; + LAYERFS_UPPERLOCK(vp, LK_EXCLUSIVE, error); + + if (error) { + /* + * How did we get a locking error? The node just came off + * of the free list, and we're the only routine which + * knows it's there... + */ + vp->v_vnlock = &vp->v_lock; + *vpp = NULL; + + /* free the substructures we've allocated. */ + FREE(xp, M_TEMP); + if (vp->v_type == VBLK || vp->v_type == VCHR) + FREE(vp->v_specinfo, M_VNODE); + + vp->v_type = VBAD; /* node is discarded */ + vp->v_op = dead_vnodeop_p; /* so ops will still work */ + vrele(vp); /* get rid of it. */ + return (error); + } + /* + * NetBSD used to do an inlined checkalias here. We do not, as + * we never flag device nodes as being aliased. The lowervp + * node will, when appropriate, be flaged as an alias. + */ + + *vpp = vp; + VREF(lowervp); /* Take into account reference held in layer_node */ + hd = LAYER_NHASH(lmp, lowervp); + LIST_INSERT_HEAD(hd, xp, layer_hash); + uvm_vnp_setsize(vp, 0); + simple_unlock(&lmp->layerm_hashlock); + return (0); +} + + +/* + * Try to find an existing layer_node vnode refering + * to it, otherwise make a new layer_node vnode which + * contains a reference to the lower vnode. + * + * >>> we assume that the lower node is already locked upon entry, so we + * propagate the lock state to upper node << + */ +int +layer_node_create(mp, lowervp, newvpp) + struct mount *mp; + struct vnode *lowervp; + struct vnode **newvpp; +{ + struct vnode *aliasvp; + struct layer_mount *lmp = MOUNTTOLAYERMOUNT(mp); + + if ((aliasvp = layer_node_find(mp, lowervp)) != NULL) { + /* + * layer_node_find has taken another reference + * to the alias vnode and moved the lock holding to + * aliasvp + */ +#ifdef LAYERFS_DIAGNOSTIC + vprint("layer_node_create: exists", aliasvp); +#endif + } else { + int error; + + /* + * Get new vnode. + */ +#ifdef LAYERFS_DIAGNOSTIC + printf("layer_node_create: create new alias vnode\n"); +#endif + + /* + * Make new vnode reference the layer_node. + */ + if ((error = (lmp->layerm_alloc)(mp, lowervp, &aliasvp)) != 0) + return error; + + /* + * aliasvp is already VREF'd by getnewvnode() + */ + } + + /* + * Now that we have VREF'd the upper vnode, release the reference + * to the lower node. The existance of the layer_node retains one + * reference to the lower node. + */ + vrele(lowervp); + +#ifdef DIAGNOSTIC + if (lowervp->v_usecount < 1) { + /* Should never happen... */ + vprint("layer_node_create: alias", aliasvp); + vprint("layer_node_create: lower", lowervp); + panic("layer_node_create: lower has 0 usecount."); + } +#endif + +#ifdef LAYERFS_DIAGNOSTIC + vprint("layer_node_create: alias", aliasvp); +#endif + *newvpp = aliasvp; + return (0); +} + +struct vnode * +layer_checkvp(vp, fil, lno) + struct vnode *vp; + char *fil; + int lno; +{ + struct layer_node *a = VTOLAYER(vp); +#ifdef notyet + /* + * Can't do this check because vop_reclaim runs + * with a funny vop vector. + * + * WRS - no it doesnt... + */ + if (vp->v_op != layer_vnodeop_p) { + printf ("layer_checkvp: on non-layer-node\n"); +#ifdef notyet + while (layer_checkvp_barrier) /*WAIT*/ ; +#endif + panic("layer_checkvp"); + } +#endif + if (a->layer_lowervp == NULL) { + /* Should never happen */ + int i; u_long *p; + printf("vp = %p, ZERO ptr\n", vp); + for (p = (u_long *) a, i = 0; i < 8; i++) + printf(" %lx", p[i]); + printf("\n"); + /* wait for debugger */ + panic("layer_checkvp"); + } + if (a->layer_lowervp->v_usecount < 1) { + int i; u_long *p; + printf("vp = %p, unref'ed lowervp\n", vp); + for (p = (u_long *) a, i = 0; i < 8; i++) + printf(" %lx", p[i]); + printf("\n"); + /* wait for debugger */ + panic ("layer with unref'ed lowervp"); + } +#ifdef notnow + printf("layer %p/%d -> %p/%d [%s, %d]\n", + LAYERTOV(a), LAYERTOV(a)->v_usecount, + a->layer_lowervp, a->layer_lowervp->v_usecount, + fil, lno); +#endif + return a->layer_lowervp; +} diff --git a/sys/miscfs/genfs/layer_vfsops.c b/sys/miscfs/genfs/layer_vfsops.c new file mode 100644 index 00000000000..68970c8ca8f --- /dev/null +++ b/sys/miscfs/genfs/layer_vfsops.c @@ -0,0 +1,285 @@ +/* $OpenBSD: layer_vfsops.c,v 1.1 2003/05/12 20:58:40 tedu Exp $ */ +/* $NetBSD: layer_vfsops.c,v 1.5 2001/11/15 09:48:21 lukem Exp $ */ + +/* + * Copyright (c) 1999 National Aeronautics & Space Administration + * All rights reserved. + * + * This software was written by William Studenmund of the + * Numerical Aerospace Simulation Facility, NASA Ames Research Center. + * + * 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. + * 3. Neither the name of the National Aeronautics & Space Administration + * nor the names of its contributors may be used to endorse or promote + * products derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY THE NATIONAL AERONAUTICS & SPACE ADMINISTRATION + * ``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 ADMINISTRATION OR CONTRIB- + * UTORS 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. + */ +/* + * Copyright (c) 1992, 1993, 1995 + * The Regents of the University of California. All rights reserved. + * + * This code is derived from software donated to Berkeley by + * Jan-Simon Pendry. + * + * 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. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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. + * + * from: Id: lofs_vfsops.c,v 1.9 1992/05/30 10:26:24 jsp Exp + * from: @(#)lofs_vfsops.c 1.2 (Berkeley) 6/18/92 + * @(#)null_vfsops.c 8.7 (Berkeley) 5/14/95 + */ + +/* + * generic layer vfs ops. + */ + + +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/time.h> +#include <sys/proc.h> +#include <sys/vnode.h> +#include <sys/mount.h> +#include <sys/namei.h> +#include <sys/malloc.h> +#include <miscfs/genfs/layer.h> +#include <miscfs/genfs/layer_extern.h> + +/* + * VFS start. Nothing needed here - the start routine + * on the underlying filesystem will have been called + * when that filesystem was mounted. + */ +int +layerfs_start(mp, flags, p) + struct mount *mp; + int flags; + struct proc *p; +{ + + return (0); + /* return VFS_START(MOUNTTOLAYERMOUNT(mp)->layerm_vfs, flags, p); */ +} + +int +layerfs_root(mp, vpp) + struct mount *mp; + struct vnode **vpp; +{ + struct vnode *vp; + +#ifdef LAYERFS_DIAGNOSTIC + printf("layerfs_root(mp = %p, vp = %p->%p)\n", mp, + MOUNTTOLAYERMOUNT(mp)->layerm_rootvp, + LAYERVPTOLOWERVP(MOUNTTOLAYERMOUNT(mp)->layerm_rootvp)); +#endif + + /* + * Return locked reference to root. + */ + vp = MOUNTTOLAYERMOUNT(mp)->layerm_rootvp; + if (vp == NULL) { + *vpp = NULL; + return (EINVAL); + } + VREF(vp); + vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, curproc); + *vpp = vp; + return 0; +} + +int +layerfs_quotactl(mp, cmd, uid, arg, p) + struct mount *mp; + int cmd; + uid_t uid; + caddr_t arg; + struct proc *p; +{ + + return VFS_QUOTACTL(MOUNTTOLAYERMOUNT(mp)->layerm_vfs, + cmd, uid, arg, p); +} + +int +layerfs_statfs(mp, sbp, p) + struct mount *mp; + struct statfs *sbp; + struct proc *p; +{ + int error; + struct statfs mstat; + +#ifdef LAYERFS_DIAGNOSTIC + printf("layerfs_statfs(mp = %p, vp = %p->%p)\n", mp, + MOUNTTOLAYERMOUNT(mp)->layerm_rootvp, + LAYERVPTOLOWERVP(MOUNTTOLAYERMOUNT(mp)->layerm_rootvp)); +#endif + + memset(&mstat, 0, sizeof(mstat)); + + error = VFS_STATFS(MOUNTTOLAYERMOUNT(mp)->layerm_vfs, &mstat, p); + if (error) + return (error); + + /* now copy across the "interesting" information and fake the rest */ + sbp->f_flags = mstat.f_flags; + sbp->f_bsize = mstat.f_bsize; + sbp->f_iosize = mstat.f_iosize; + sbp->f_blocks = mstat.f_blocks; + sbp->f_bfree = mstat.f_bfree; + sbp->f_bavail = mstat.f_bavail; + sbp->f_files = mstat.f_files; + sbp->f_ffree = mstat.f_ffree; + if (sbp != &mp->mnt_stat) { + memcpy(&sbp->f_fsid, &mp->mnt_stat.f_fsid, sizeof(sbp->f_fsid)); + memcpy(sbp->f_mntonname, mp->mnt_stat.f_mntonname, MNAMELEN); + memcpy(sbp->f_mntfromname, mp->mnt_stat.f_mntfromname, MNAMELEN); + } + strncpy(sbp->f_fstypename, mp->mnt_vfc->vfc_name, MFSNAMELEN); + return (0); +} + +int +layerfs_sync(mp, waitfor, cred, p) + struct mount *mp; + int waitfor; + struct ucred *cred; + struct proc *p; +{ + + /* + * XXX - Assumes no data cached at layer. + */ + return (0); +} + +int +layerfs_vget(mp, ino, vpp) + struct mount *mp; + ino_t ino; + struct vnode **vpp; +{ + int error; + struct vnode *vp; + + if ((error = VFS_VGET(MOUNTTOLAYERMOUNT(mp)->layerm_vfs, ino, &vp))) { + *vpp = NULL; + return (error); + } + if ((error = layer_node_create(mp, vp, vpp))) { + vput(vp); + *vpp = NULL; + return (error); + } + + return (0); +} + +int +layerfs_fhtovp(mp, fidp, vpp) + struct mount *mp; + struct fid *fidp; + struct vnode **vpp; +{ + int error; + struct vnode *vp; + + if ((error = VFS_FHTOVP(MOUNTTOLAYERMOUNT(mp)->layerm_vfs, fidp, &vp))) + return (error); + + if ((error = layer_node_create(mp, vp, vpp))) { + vput(vp); + *vpp = NULL; + return (error); + } + + return (0); +} + +int +layerfs_checkexp(mp, nam, exflagsp, credanonp) + struct mount *mp; + struct mbuf *nam; + int *exflagsp; + struct ucred**credanonp; +{ + struct netcred *np; + struct layer_mount *lmp = MOUNTTOLAYERMOUNT(mp); + + /* + * get the export permission structure for this <mp, client> tuple. + */ + if ((np = vfs_export_lookup(mp, &lmp->layerm_export, nam)) == NULL) + return (EACCES); + + *exflagsp = np->netc_exflags; + *credanonp = &np->netc_anon; + return (0); +} + +int +layerfs_vptofh(vp, fhp) + struct vnode *vp; + struct fid *fhp; +{ + + return (VFS_VPTOFH(LAYERVPTOLOWERVP(vp), fhp)); +} + +int +layerfs_sysctl(name, namelen, oldp, oldlenp, newp, newlen, p) + int *name; + u_int namelen; + void *oldp; + size_t *oldlenp; + void *newp; + size_t newlen; + struct proc *p; +{ + return (EOPNOTSUPP); +} diff --git a/sys/miscfs/genfs/layer_vnops.c b/sys/miscfs/genfs/layer_vnops.c new file mode 100644 index 00000000000..2b644098d09 --- /dev/null +++ b/sys/miscfs/genfs/layer_vnops.c @@ -0,0 +1,950 @@ +/* $OpenBSD: layer_vnops.c,v 1.1 2003/05/12 20:58:40 tedu Exp $ */ +/* $NetBSD: layer_vnops.c,v 1.10 2001/12/06 04:29:23 chs Exp $ */ + +/* + * Copyright (c) 1999 National Aeronautics & Space Administration + * All rights reserved. + * + * This software was written by William Studenmund of the + * Numerical Aerospace Simulation Facility, NASA Ames Research Center. + * + * 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. + * 3. Neither the name of the National Aeronautics & Space Administration + * nor the names of its contributors may be used to endorse or promote + * products derived from this software without specific prior written + * permission. + * + * THIS SOFTWARE IS PROVIDED BY THE NATIONAL AERONAUTICS & SPACE ADMINISTRATION + * ``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 ADMINISTRATION OR CONTRIB- + * UTORS 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. + */ +/* + * Copyright (c) 1992, 1993 + * The Regents of the University of California. All rights reserved. + * + * This code is derived from software contributed to Berkeley by + * John Heidemann of the UCLA Ficus project. + * + * 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. + * 3. All advertising materials mentioning features or use of this software + * must display the following acknowledgement: + * This product includes software developed by the University of + * California, Berkeley and its contributors. + * 4. Neither the name of the University nor the names of its contributors + * may be used to endorse or promote products derived from this software + * without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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. + * + * @(#)null_vnops.c 8.6 (Berkeley) 5/27/95 + * + * Ancestors: + * @(#)lofs_vnops.c 1.2 (Berkeley) 6/18/92 + * $Id: layer_vnops.c,v 1.1 2003/05/12 20:58:40 tedu Exp $ + * ...and... + * @(#)null_vnodeops.c 1.20 92/07/07 UCLA Ficus project + */ + +/* + * Null Layer vnode routines. + * + * (See mount_null(8) for more information.) + * + * The layer.h, layer_extern.h, layer_vfs.c, and layer_vnops.c files provide + * the core implimentation of the null file system and most other stacked + * fs's. The description below refers to the null file system, but the + * services provided by the layer* files are useful for all layered fs's. + * + * The null layer duplicates a portion of the file system + * name space under a new name. In this respect, it is + * similar to the loopback file system. It differs from + * the loopback fs in two respects: it is implemented using + * a stackable layers techniques, and it's "null-node"s stack above + * all lower-layer vnodes, not just over directory vnodes. + * + * The null layer has two purposes. First, it serves as a demonstration + * of layering by proving a layer which does nothing. (It actually + * does everything the loopback file system does, which is slightly + * more than nothing.) Second, the null layer can serve as a prototype + * layer. Since it provides all necessary layer framework, + * new file system layers can be created very easily be starting + * with a null layer. + * + * The remainder of the man page examines the null layer as a basis + * for constructing new layers. + * + * + * INSTANTIATING NEW NULL LAYERS + * + * New null layers are created with mount_null(8). + * Mount_null(8) takes two arguments, the pathname + * of the lower vfs (target-pn) and the pathname where the null + * layer will appear in the namespace (alias-pn). After + * the null layer is put into place, the contents + * of target-pn subtree will be aliased under alias-pn. + * + * It is conceivable that other overlay filesystems will take different + * parameters. For instance, data migration or access controll layers might + * only take one pathname which will serve both as the target-pn and + * alias-pn described above. + * + * + * OPERATION OF A NULL LAYER + * + * The null layer is the minimum file system layer, + * simply bypassing all possible operations to the lower layer + * for processing there. The majority of its activity centers + * on the bypass routine, though which nearly all vnode operations + * pass. + * + * The bypass routine accepts arbitrary vnode operations for + * handling by the lower layer. It begins by examing vnode + * operation arguments and replacing any layered nodes by their + * lower-layer equivlants. It then invokes the operation + * on the lower layer. Finally, it replaces the layered nodes + * in the arguments and, if a vnode is return by the operation, + * stacks a layered node on top of the returned vnode. + * + * The bypass routine in this file, layer_bypass(), is suitable for use + * by many different layered filesystems. It can be used by multiple + * filesystems simultaneously. Alternatively, a layered fs may provide + * its own bypass routine, in which case layer_bypass() should be used as + * a model. For instance, the main functionality provided by umapfs, the user + * identity mapping file system, is handled by a custom bypass routine. + * + * Typically a layered fs registers its selected bypass routine as the + * default vnode operation in its vnodeopv_entry_desc table. Additionally + * the filesystem must store the bypass entry point in the layerm_bypass + * field of struct layer_mount. All other layer routines in this file will + * use the layerm_bypass routine. + * + * Although the bypass routine handles most operations outright, a number + * of operations are special cased, and handled by the layered fs. One + * group, layer_setattr, layer_getattr, layer_access, layer_open, and + * layer_fsync, perform layer-specific manipulation in addition to calling + * the bypass routine. The other group + + * Although bypass handles most operations, vop_getattr, vop_lock, + * vop_unlock, vop_inactive, vop_reclaim, and vop_print are not + * bypassed. Vop_getattr must change the fsid being returned. + * Vop_lock and vop_unlock must handle any locking for the + * current vnode as well as pass the lock request down. + * Vop_inactive and vop_reclaim are not bypassed so that + * they can handle freeing null-layer specific data. Vop_print + * is not bypassed to avoid excessive debugging information. + * Also, certain vnode operations change the locking state within + * the operation (create, mknod, remove, link, rename, mkdir, rmdir, + * and symlink). Ideally these operations should not change the + * lock state, but should be changed to let the caller of the + * function unlock them. Otherwise all intermediate vnode layers + * (such as union, umapfs, etc) must catch these functions to do + * the necessary locking at their layer. + * + * + * INSTANTIATING VNODE STACKS + * + * Mounting associates the null layer with a lower layer, + * effect stacking two VFSes. Vnode stacks are instead + * created on demand as files are accessed. + * + * The initial mount creates a single vnode stack for the + * root of the new null layer. All other vnode stacks + * are created as a result of vnode operations on + * this or other null vnode stacks. + * + * New vnode stacks come into existance as a result of + * an operation which returns a vnode. + * The bypass routine stacks a null-node above the new + * vnode before returning it to the caller. + * + * For example, imagine mounting a null layer with + * "mount_null /usr/include /dev/layer/null". + * Changing directory to /dev/layer/null will assign + * the root null-node (which was created when the null layer was mounted). + * Now consider opening "sys". A vop_lookup would be + * done on the root null-node. This operation would bypass through + * to the lower layer which would return a vnode representing + * the UFS "sys". layer_bypass then builds a null-node + * aliasing the UFS "sys" and returns this to the caller. + * Later operations on the null-node "sys" will repeat this + * process when constructing other vnode stacks. + * + * + * CREATING OTHER FILE SYSTEM LAYERS + * + * One of the easiest ways to construct new file system layers is to make + * a copy of the null layer, rename all files and variables, and + * then begin modifing the copy. Sed can be used to easily rename + * all variables. + * + * The umap layer is an example of a layer descended from the + * null layer. + * + * + * INVOKING OPERATIONS ON LOWER LAYERS + * + * There are two techniques to invoke operations on a lower layer + * when the operation cannot be completely bypassed. Each method + * is appropriate in different situations. In both cases, + * it is the responsibility of the aliasing layer to make + * the operation arguments "correct" for the lower layer + * by mapping an vnode arguments to the lower layer. + * + * The first approach is to call the aliasing layer's bypass routine. + * This method is most suitable when you wish to invoke the operation + * currently being hanldled on the lower layer. It has the advantage + * that the bypass routine already must do argument mapping. + * An example of this is null_getattrs in the null layer. + * + * A second approach is to directly invoked vnode operations on + * the lower layer with the VOP_OPERATIONNAME interface. + * The advantage of this method is that it is easy to invoke + * arbitrary operations on the lower layer. The disadvantage + * is that vnodes arguments must be manualy mapped. + * + */ + + +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/proc.h> +#include <sys/time.h> +#include <sys/vnode.h> +#include <sys/mount.h> +#include <sys/namei.h> +#include <sys/malloc.h> +#include <sys/buf.h> +#include <miscfs/genfs/layer.h> +#include <miscfs/genfs/layer_extern.h> + + +/* + * This is the 08-June-99 bypass routine, based on the 10-Apr-92 bypass + * routine by John Heidemann. + * The new element for this version is that the whole nullfs + * system gained the concept of locks on the lower node, and locks on + * our nodes. When returning from a call to the lower layer, we may + * need to update lock state ONLY on our layer. The LAYERFS_UPPER*LOCK() + * macros provide this functionality. + * The 10-Apr-92 version was optimized for speed, throwing away some + * safety checks. It should still always work, but it's not as + * robust to programmer errors. + * Define SAFETY to include some error checking code. + * + * In general, we map all vnodes going down and unmap them on the way back. + * + * Also, some BSD vnode operations have the side effect of vrele'ing + * their arguments. With stacking, the reference counts are held + * by the upper node, not the lower one, so we must handle these + * side-effects here. This is not of concern in Sun-derived systems + * since there are no such side-effects. + * + * New for the 08-June-99 version: we also handle operations which unlock + * the passed-in node (typically they vput the node). + * + * This makes the following assumptions: + * - only one returned vpp + * - no INOUT vpp's (Sun's vop_open has one of these) + * - the vnode operation vector of the first vnode should be used + * to determine what implementation of the op should be invoked + * - all mapped vnodes are of our vnode-type (NEEDSWORK: + * problems on rmdir'ing mount points and renaming?) + */ +int +layer_bypass(v) + void *v; +{ + struct vop_generic_args /* { + struct vnodeop_desc *a_desc; + <other random data follows, presumably> + } */ *ap = v; + int (**our_vnodeop_p)(void *); + struct vnode **this_vp_p; + int error, error1; + struct vnode *old_vps[VDESC_MAX_VPS], *vp0; + struct vnode **vps_p[VDESC_MAX_VPS]; + struct vnode ***vppp; + struct vnodeop_desc *descp = ap->a_desc; + int reles, i, flags; + +#ifdef SAFETY + /* + * We require at least one vp. + */ + if (descp->vdesc_vp_offsets == NULL || + descp->vdesc_vp_offsets[0] == VDESC_NO_OFFSET) + panic ("layer_bypass: no vp's in map.\n"); +#endif + + vps_p[0] = VOPARG_OFFSETTO(struct vnode**,descp->vdesc_vp_offsets[0],ap); + vp0 = *vps_p[0]; + flags = MOUNTTOLAYERMOUNT(vp0->v_mount)->layerm_flags; + our_vnodeop_p = vp0->v_op; + + if (flags & LAYERFS_MBYPASSDEBUG) + printf ("layer_bypass: %s\n", descp->vdesc_name); + + /* + * Map the vnodes going in. + * Later, we'll invoke the operation based on + * the first mapped vnode's operation vector. + */ + reles = descp->vdesc_flags; + for (i = 0; i < VDESC_MAX_VPS; reles >>= 1, i++) { + if (descp->vdesc_vp_offsets[i] == VDESC_NO_OFFSET) + break; /* bail out at end of list */ + vps_p[i] = this_vp_p = + VOPARG_OFFSETTO(struct vnode**,descp->vdesc_vp_offsets[i],ap); + /* + * We're not guaranteed that any but the first vnode + * are of our type. Check for and don't map any + * that aren't. (We must always map first vp or vclean fails.) + */ + if (i && (*this_vp_p == NULL || + (*this_vp_p)->v_op != our_vnodeop_p)) { + old_vps[i] = NULL; + } else { + old_vps[i] = *this_vp_p; + *(vps_p[i]) = LAYERVPTOLOWERVP(*this_vp_p); + /* + * XXX - Several operations have the side effect + * of vrele'ing their vp's. We must account for + * that. (This should go away in the future.) + */ + if (reles & VDESC_VP0_WILLRELE) + VREF(*this_vp_p); + } + + } + + /* + * Call the operation on the lower layer + * with the modified argument structure. + */ + error = VCALL(*vps_p[0], descp->vdesc_offset, ap); + + /* + * Maintain the illusion of call-by-value + * by restoring vnodes in the argument structure + * to their original value. + */ + reles = descp->vdesc_flags; + for (i = 0; i < VDESC_MAX_VPS; reles >>= 1, i++) { + if (descp->vdesc_vp_offsets[i] == VDESC_NO_OFFSET) + break; /* bail out at end of list */ + if (old_vps[i]) { + *(vps_p[i]) = old_vps[i]; + if (reles & VDESC_VP0_WILLUNLOCK) + LAYERFS_UPPERUNLOCK(*(vps_p[i]), 0, error1); + if (reles & VDESC_VP0_WILLRELE) + vrele(*(vps_p[i])); + } + } + + /* + * Map the possible out-going vpp + * (Assumes that the lower layer always returns + * a VREF'ed vpp unless it gets an error.) + */ + if (descp->vdesc_vpp_offset != VDESC_NO_OFFSET && + !(descp->vdesc_flags & VDESC_NOMAP_VPP) && + !error) { + /* + * XXX - even though some ops have vpp returned vp's, + * several ops actually vrele this before returning. + * We must avoid these ops. + * (This should go away when these ops are regularized.) + */ + if (descp->vdesc_flags & VDESC_VPP_WILLRELE) + goto out; + vppp = VOPARG_OFFSETTO(struct vnode***, + descp->vdesc_vpp_offset,ap); + /* + * Only vop_lookup, vop_create, vop_makedir, vop_bmap, + * vop_mknod, and vop_symlink return vpp's. vop_bmap + * doesn't call bypass as the lower vpp is fine (we're just + * going to do i/o on it). vop_loookup doesn't call bypass + * as a lookup on "." would generate a locking error. + * So all the calls which get us here have a locked vpp. :-) + */ + error = layer_node_create(old_vps[0]->v_mount, **vppp, *vppp); + } + + out: + return (error); +} + +/* + * We have to carry on the locking protocol on the layer vnodes + * as we progress through the tree. We also have to enforce read-only + * if this layer is mounted read-only. + */ +int +layer_lookup(v) + void *v; +{ + struct vop_lookup_args /* { + struct vnodeop_desc *a_desc; + struct vnode * a_dvp; + struct vnode ** a_vpp; + struct componentname * a_cnp; + } */ *ap = v; + struct componentname *cnp = ap->a_cnp; + int flags = cnp->cn_flags; + struct vnode *dvp, *vp, *ldvp; + int error, r; + + dvp = ap->a_dvp; + *ap->a_vpp = NULL; + + if ((flags & ISLASTCN) && (dvp->v_mount->mnt_flag & MNT_RDONLY) && + (cnp->cn_nameiop == DELETE || cnp->cn_nameiop == RENAME)) + return (EROFS); + + ldvp = LAYERVPTOLOWERVP(dvp); + ap->a_dvp = ldvp; + error = VCALL(ldvp, ap->a_desc->vdesc_offset, ap); + vp = *ap->a_vpp; + + if (error == EJUSTRETURN && (flags & ISLASTCN) && + (dvp->v_mount->mnt_flag & MNT_RDONLY) && + (cnp->cn_nameiop == CREATE || cnp->cn_nameiop == RENAME)) + error = EROFS; + /* + * We must do the same locking and unlocking at this layer as + * is done in the layers below us. It used to be we would try + * to guess based on what was set with the flags and error codes. + * + * But that doesn't work. So now we have the underlying VOP_LOOKUP + * tell us if it released the parent vnode, and we adjust the + * upper node accordingly. We can't just look at the lock states + * of the lower nodes as someone else might have come along and + * locked the parent node after our call to VOP_LOOKUP locked it. + */ + if ((cnp->cn_flags & PDIRUNLOCK)) { + LAYERFS_UPPERUNLOCK(dvp, 0, r); + } + if (ldvp == vp) { + /* + * Did lookup on "." or ".." in the root node of a mount point. + * So we return dvp after a VREF. + */ + *ap->a_vpp = dvp; + VREF(dvp); + vrele(vp); + } else if (vp != NULL) { + error = layer_node_create(dvp->v_mount, vp, ap->a_vpp); + } + return (error); +} + +/* + * Setattr call. Disallow write attempts if the layer is mounted read-only. + */ +int +layer_setattr(v) + void *v; +{ + struct vop_setattr_args /* { + struct vnodeop_desc *a_desc; + struct vnode *a_vp; + struct vattr *a_vap; + struct ucred *a_cred; + struct proc *a_p; + } */ *ap = v; + struct vnode *vp = ap->a_vp; + struct vattr *vap = ap->a_vap; + + if ((vap->va_flags != VNOVAL || vap->va_uid != (uid_t)VNOVAL || + vap->va_gid != (gid_t)VNOVAL || vap->va_atime.tv_sec != VNOVAL || + vap->va_mtime.tv_sec != VNOVAL || vap->va_mode != (mode_t)VNOVAL) && + (vp->v_mount->mnt_flag & MNT_RDONLY)) + return (EROFS); + if (vap->va_size != VNOVAL) { + switch (vp->v_type) { + case VDIR: + return (EISDIR); + case VCHR: + case VBLK: + case VSOCK: + case VFIFO: + return (0); + case VREG: + case VLNK: + default: + /* + * Disallow write attempts if the filesystem is + * mounted read-only. + */ + if (vp->v_mount->mnt_flag & MNT_RDONLY) + return (EROFS); + } + } + return (LAYERFS_DO_BYPASS(vp, ap)); +} + +/* + * We handle getattr only to change the fsid. + */ +int +layer_getattr(v) + void *v; +{ + struct vop_getattr_args /* { + struct vnode *a_vp; + struct vattr *a_vap; + struct ucred *a_cred; + struct proc *a_p; + } */ *ap = v; + struct vnode *vp = ap->a_vp; + int error; + + if ((error = LAYERFS_DO_BYPASS(vp, ap)) != 0) + return (error); + /* Requires that arguments be restored. */ + ap->a_vap->va_fsid = vp->v_mount->mnt_stat.f_fsid.val[0]; + return (0); +} + +int +layer_access(v) + void *v; +{ + struct vop_access_args /* { + struct vnode *a_vp; + int a_mode; + struct ucred *a_cred; + struct proc *a_p; + } */ *ap = v; + struct vnode *vp = ap->a_vp; + mode_t mode = ap->a_mode; + + /* + * Disallow write attempts on read-only layers; + * unless the file is a socket, fifo, or a block or + * character device resident on the file system. + */ + if (mode & VWRITE) { + switch (vp->v_type) { + case VDIR: + case VLNK: + case VREG: + if (vp->v_mount->mnt_flag & MNT_RDONLY) + return (EROFS); + break; + default: + break; + } + } + return (LAYERFS_DO_BYPASS(vp, ap)); +} + +/* + * We must handle open to be able to catch MNT_NODEV and friends. + */ +int +layer_open(v) + void *v; +{ + struct vop_open_args *ap = v; + struct vnode *vp = ap->a_vp; + enum vtype lower_type = LAYERVPTOLOWERVP(vp)->v_type; + + if (((lower_type == VBLK) || (lower_type == VCHR)) && + (vp->v_mount->mnt_flag & MNT_NODEV)) + return ENXIO; + + return LAYERFS_DO_BYPASS(vp, ap); +} + +/* + * We need to process our own vnode lock and then clear the + * interlock flag as it applies only to our vnode, not the + * vnodes below us on the stack. + */ +int +layer_lock(v) + void *v; +{ + struct vop_lock_args /* { + struct vnode *a_vp; + int a_flags; + struct proc *a_p; + } */ *ap = v; + struct vnode *vp = ap->a_vp, *lowervp; + int flags = ap->a_flags, error; + struct proc *p = ap->a_p; + + if (vp->v_flag & VXLOCK) /* XXX this is a disaster -tedu XXX */ + return (0); + + if (vp->v_vnlock != NULL) { + /* + * The lower level has exported a struct lock to us. Use + * it so that all vnodes in the stack lock and unlock + * simultaneously. Note: we don't DRAIN the lock as DRAIN + * decommissions the lock - just because our vnode is + * going away doesn't mean the struct lock below us is. + * LK_EXCLUSIVE is fine. + */ + if ((flags & LK_TYPE_MASK) == LK_DRAIN) + return (lockmgr(vp->v_vnlock, (flags & ~LK_TYPE_MASK) | + LK_EXCLUSIVE, &vp->v_interlock, p)); + else + return (lockmgr(vp->v_vnlock, flags, &vp->v_interlock, p)); + } else { + /* + * Ahh well. It would be nice if the fs we're over would + * export a struct lock for us to use, but it doesn't. + * + * To prevent race conditions involving doing a lookup + * on "..", we have to lock the lower node, then lock our + * node. Most of the time it won't matter that we lock our + * node (as any locking would need the lower one locked + * first). But we can LK_DRAIN the upper lock as a step + * towards decomissioning it. + */ + lowervp = LAYERVPTOLOWERVP(vp); + if (flags & LK_INTERLOCK) { + simple_unlock(&vp->v_interlock); + flags &= ~LK_INTERLOCK; + } + if ((flags & LK_TYPE_MASK) == LK_DRAIN) { + error = VOP_LOCK(lowervp, (flags & ~LK_TYPE_MASK) | + LK_EXCLUSIVE, p); + } else { + error = VOP_LOCK(lowervp, flags, p); + } + if (error) + return (error); + if ((error = lockmgr(&vp->v_lock, flags, &vp->v_interlock, p))) { + VOP_UNLOCK(lowervp, 0, p); + } + return (error); + } +} + +/* + */ +int +layer_unlock(v) + void *v; +{ + struct vop_unlock_args /* { + struct vnode *a_vp; + int a_flags; + struct proc *a_p; + } */ *ap = v; + struct vnode *vp = ap->a_vp; + int flags = ap->a_flags; + struct proc *p = ap->a_p; + + if (vp->v_flag & VXLOCK) + return (0); + + if (vp->v_vnlock != NULL) { + return (lockmgr(vp->v_vnlock, ap->a_flags | LK_RELEASE, + &vp->v_interlock, p)); + } else { + if (flags & LK_INTERLOCK) { + simple_unlock(&vp->v_interlock); + flags &= ~LK_INTERLOCK; + } + VOP_UNLOCK(LAYERVPTOLOWERVP(vp), flags, p); + return (lockmgr(&vp->v_lock, ap->a_flags | LK_RELEASE, + &vp->v_interlock, p)); + } +} + +/* + * As long as genfs_nolock is in use, don't call VOP_ISLOCKED(lowervp) + * if vp->v_vnlock == NULL as genfs_noislocked will always report 0. + */ +int +layer_islocked(v) + void *v; +{ + struct vop_islocked_args /* { + struct vnode *a_vp; + } */ *ap = v; + struct vnode *vp = ap->a_vp; + + if (vp->v_vnlock != NULL) + return (lockstatus(vp->v_vnlock)); + else + return (lockstatus(&vp->v_lock)); +} + +/* + * If vinvalbuf is calling us, it's a "shallow fsync" -- don't bother + * syncing the underlying vnodes, since they'll be fsync'ed when + * reclaimed; otherwise, + * pass it through to the underlying layer. + * + * XXX Do we still need to worry about shallow fsync? + */ + +int +layer_fsync(v) + void *v; +{ + struct vop_fsync_args /* { + struct vnode *a_vp; + struct ucred *a_cred; + int a_flags; + off_t offlo; + off_t offhi; + struct proc *a_p; + } */ *ap = v; + + return (LAYERFS_DO_BYPASS(ap->a_vp, ap)); +} + + +int +layer_inactive(v) + void *v; +{ + struct vop_inactive_args /* { + struct vnode *a_vp; + struct proc *a_p; + } */ *ap = v; + struct vnode *vp = ap->a_vp; + struct proc *p = ap->a_p; + + /* + * Do nothing (and _don't_ bypass). + * Wait to vrele lowervp until reclaim, + * so that until then our layer_node is in the + * cache and reusable. + * + * NEEDSWORK: Someday, consider inactive'ing + * the lowervp and then trying to reactivate it + * with capabilities (v_id) + * like they do in the name lookup cache code. + * That's too much work for now. + */ + VOP_UNLOCK(vp, 0, p); + + /* ..., but don't cache the device node. */ + if (vp->v_type == VBLK || vp->v_type == VCHR) + vgone(vp); + return (0); +} + +int +layer_reclaim(v) + void *v; +{ + struct vop_reclaim_args /* { + struct vnode *a_vp; + struct proc *a_p; + } */ *ap = v; + struct vnode *vp = ap->a_vp; + struct layer_mount *lmp = MOUNTTOLAYERMOUNT(vp->v_mount); + struct layer_node *xp = VTOLAYER(vp); + struct vnode *lowervp = xp->layer_lowervp; + + /* + * Note: in vop_reclaim, the node's struct lock has been + * decomissioned, so we have to be careful about calling + * VOP's on ourself. Even if we turned a LK_DRAIN into an + * LK_EXCLUSIVE in layer_lock, we still must be careful as VXLOCK is + * set. + */ + /* After this assignment, this node will not be re-used. */ + if ((vp == lmp->layerm_rootvp)) { + /* + * Oops! We no longer have a root node. Most likely reason is + * that someone forcably unmunted the underlying fs. + * + * Now getting the root vnode will fail. We're dead. :-( + */ + lmp->layerm_rootvp = NULL; + } + xp->layer_lowervp = NULL; + simple_lock(&lmp->layerm_hashlock); + LIST_REMOVE(xp, layer_hash); + simple_unlock(&lmp->layerm_hashlock); + FREE(vp->v_data, M_TEMP); + vp->v_data = NULL; + vrele (lowervp); + return (0); +} + +/* + * We just feed the returned vnode up to the caller - there's no need + * to build a layer node on top of the node on which we're going to do + * i/o. :-) + */ +int +layer_bmap(v) + void *v; +{ + struct vop_bmap_args /* { + struct vnode *a_vp; + daddr_t a_bn; + struct vnode **a_vpp; + daddr_t *a_bnp; + int *a_runp; + } */ *ap = v; + struct vnode *vp; + + ap->a_vp = vp = LAYERVPTOLOWERVP(ap->a_vp); + + return (VCALL(vp, ap->a_desc->vdesc_offset, ap)); +} + +int +layer_print(v) + void *v; +{ + struct vop_print_args /* { + struct vnode *a_vp; + } */ *ap = v; + struct vnode *vp = ap->a_vp; + printf ("\ttag VT_LAYERFS, vp=%p, lowervp=%p\n", vp, LAYERVPTOLOWERVP(vp)); + VOP_PRINT(LAYERVPTOLOWERVP(vp)); + return (0); +} + +/* + * XXX - vop_strategy must be hand coded because it has no + * vnode in its arguments. + * This goes away with a merged VM/buffer cache. + */ +int +layer_strategy(v) + void *v; +{ + struct vop_strategy_args /* { + struct buf *a_bp; + } */ *ap = v; + struct buf *bp = ap->a_bp; + int error; + struct vnode *savedvp; + + savedvp = bp->b_vp; + bp->b_vp = LAYERVPTOLOWERVP(bp->b_vp); + + error = VOP_STRATEGY(bp); + + bp->b_vp = savedvp; + + return (error); +} + +/* + * XXX - like vop_strategy, vop_bwrite must be hand coded because it has no + * vnode in its arguments. + * This goes away with a merged VM/buffer cache. + */ +int +layer_bwrite(v) + void *v; +{ + struct vop_bwrite_args /* { + struct buf *a_bp; + } */ *ap = v; + struct buf *bp = ap->a_bp; + int error; + struct vnode *savedvp; + + savedvp = bp->b_vp; + bp->b_vp = LAYERVPTOLOWERVP(bp->b_vp); + + error = VOP_BWRITE(bp); + + bp->b_vp = savedvp; + + return (error); +} + +#if 0 +int +layer_getpages(v) + void *v; +{ + struct vop_getpages_args /* { + struct vnode *a_vp; + voff_t a_offset; + struct vm_page **a_m; + int *a_count; + int a_centeridx; + vm_prot_t a_access_type; + int a_advice; + int a_flags; + } */ *ap = v; + struct vnode *vp = ap->a_vp; + int error; + + /* + * just pass the request on to the underlying layer. + */ + + if (ap->a_flags & PGO_LOCKED) { + return EBUSY; + } + ap->a_vp = LAYERVPTOLOWERVP(vp); + simple_unlock(&vp->v_interlock); + simple_lock(&ap->a_vp->v_interlock); + error = VCALL(ap->a_vp, VOFFSET(vop_getpages), ap); + return error; +} + +int +layer_putpages(v) + void *v; +{ + struct vop_putpages_args /* { + struct vnode *a_vp; + voff_t a_offlo; + voff_t a_offhi; + int a_flags; + } */ *ap = v; + struct vnode *vp = ap->a_vp; + int error; + + /* + * just pass the request on to the underlying layer. + */ + + ap->a_vp = LAYERVPTOLOWERVP(vp); + simple_unlock(&vp->v_interlock); + simple_lock(&ap->a_vp->v_interlock); + error = VCALL(ap->a_vp, VOFFSET(vop_putpages), ap); + return error; +} +#endif |