bootfs_vfsops.c revision 76ca3cb000306bc4052fe820a7e4a6998dbcf932
/*
* This file and its contents are supplied under the terms of the
* Common Development and Distribution License ("CDDL"), version 1.0.
* You may only use this file in accordance with the terms of version
* 1.0 of the CDDL.
*
* A full copy of the text of the CDDL should have accompanied this
* source. A copy of the CDDL is also available via the Internet at
* http://www.illumos.org/license/CDDL.
*/
/*
* Copyright (c) 2015 Joyent, Inc.
*/
#include <sys/errno.h>
#include <sys/modctl.h>
#include <sys/types.h>
#include <sys/mkdev.h>
#include <sys/ddi.h>
#include <sys/sunddi.h>
#include <sys/vfs.h>
#include <sys/vfs_opreg.h>
#include <sys/systm.h>
#include <sys/id_space.h>
#include <sys/cmn_err.h>
#include <sys/ksynch.h>
#include <sys/policy.h>
#include <sys/mount.h>
#include <sys/sysmacros.h>
#include <sys/fs/bootfs_impl.h>
/*
* While booting, additional types of modules and files can be passed in to the
* loader. These include the familiar boot archive, as well as, a module hash
* and additional modules that are interpreted as files. As part of the handoff
* in early boot, information about these modules are saved as properties on the
* root of the devinfo tree, similar to other boot-time properties.
*
* This file system provides a read-only view of those additional files. Due to
* its limited scope, it has a slightly simpler construction than several other
* file systems. When mounted, it looks for the corresponding properties and
* creates bootfs_node_t's and vnodes for all of the corresponding files and
* directories that exist along the way. At this time, there are currently a
* rather small number of files passed in this way.
*
* This does lead to one behavior that folks used to other file systems might
* find peculiar. Because we are not always actively creating and destroying the
* required vnodes on demand, the count on the root vnode will not be going up
* accordingly with the existence of other vnodes. This means that a bootfs file
* system that is not in use will have all of its vnodes exist with a v_count of
* one.
*/
major_t bootfs_major;
static int bootfs_fstype;
static id_space_t *bootfs_idspace;
static uint64_t bootfs_nactive;
static kmutex_t bootfs_lock;
static const char *bootfs_name = "bootfs";
static int
bootfs_mount(vfs_t *vfsp, vnode_t *mvp, struct mounta *uap, cred_t *cr)
{
int ret;
bootfs_t *bfs;
struct pathname dpn;
dev_t fsdev;
if ((ret = secpolicy_fs_mount(cr, mvp, vfsp)) != 0)
return (ret);
if (mvp->v_type != VDIR)
return (ENOTDIR);
if (uap->flags & MS_REMOUNT)
return (EBUSY);
mutex_enter(&mvp->v_lock);
if ((uap->flags & MS_OVERLAY) == 0 &&
(mvp->v_count != 1 || (mvp->v_flag & VROOT))) {
mutex_exit(&mvp->v_lock);
return (EBUSY);
}
mutex_exit(&mvp->v_lock);
/*
* We indicate that the backing store is bootfs. We don't want to use
* swap, because folks might think that this is putting all the data
* into memory ala tmpfs. Rather these modules are always in memory and
* there's nothing to be done about that.
*/
vfs_setresource(vfsp, bootfs_name, 0);
bfs = kmem_zalloc(sizeof (bootfs_t), KM_NOSLEEP | KM_NORMALPRI);
if (bfs == NULL)
return (ENOMEM);
ret = pn_get(uap->dir,
(uap->flags & MS_SYSSPACE) ? UIO_SYSSPACE : UIO_USERSPACE, &dpn);
if (ret != 0) {
kmem_free(bfs, sizeof (bfs));
return (ret);
}
bfs->bfs_minor = id_alloc(bootfs_idspace);
bfs->bfs_kstat = kstat_create_zone("bootfs", bfs->bfs_minor, "bootfs",
"fs", KSTAT_TYPE_NAMED,
sizeof (bootfs_stat_t) / sizeof (kstat_named_t),
KSTAT_FLAG_VIRTUAL, GLOBAL_ZONEID);
if (bfs->bfs_kstat == NULL) {
id_free(bootfs_idspace, bfs->bfs_minor);
pn_free(&dpn);
kmem_free(bfs, sizeof (bfs));
return (ENOMEM);
}
bfs->bfs_kstat->ks_data = &bfs->bfs_stat;
fsdev = makedevice(bootfs_major, bfs->bfs_minor);
bfs->bfs_vfsp = vfsp;
vfsp->vfs_data = (caddr_t)bfs;
vfsp->vfs_fstype = bootfs_fstype;
vfsp->vfs_dev = fsdev;
vfsp->vfs_bsize = PAGESIZE;
vfsp->vfs_flag |= VFS_RDONLY | VFS_NOSETUID | VFS_NOTRUNC |
VFS_UNLINKABLE;
vfs_make_fsid(&vfsp->vfs_fsid, fsdev, bootfs_fstype);
bfs->bfs_mntpath = kmem_alloc(dpn.pn_pathlen + 1, KM_SLEEP);
bcopy(dpn.pn_path, bfs->bfs_mntpath, dpn.pn_pathlen);
bfs->bfs_mntpath[dpn.pn_pathlen] = '\0';
pn_free(&dpn);
list_create(&bfs->bfs_nodes, sizeof (bootfs_node_t),
offsetof(bootfs_node_t, bvn_alink));
kstat_named_init(&bfs->bfs_stat.bfss_nfiles, "nfiles",
KSTAT_DATA_UINT32);
kstat_named_init(&bfs->bfs_stat.bfss_ndirs, "ndirs",
KSTAT_DATA_UINT32);
kstat_named_init(&bfs->bfs_stat.bfss_nbytes, "nbytes",
KSTAT_DATA_UINT64);
kstat_named_init(&bfs->bfs_stat.bfss_ndups, "ndup",
KSTAT_DATA_UINT32);
kstat_named_init(&bfs->bfs_stat.bfss_ndiscards, "ndiscard",
KSTAT_DATA_UINT32);
bootfs_construct(bfs);
kstat_install(bfs->bfs_kstat);
return (0);
}
static int
bootfs_unmount(vfs_t *vfsp, int flag, cred_t *cr)
{
int ret;
bootfs_t *bfs = vfsp->vfs_data;
bootfs_node_t *bnp;
if ((ret = secpolicy_fs_unmount(cr, vfsp)) != 0)
return (ret);
if (flag & MS_FORCE)
return (ENOTSUP);
for (bnp = list_head(&bfs->bfs_nodes); bnp != NULL;
bnp = list_next(&bfs->bfs_nodes, bnp)) {
mutex_enter(&bnp->bvn_vnp->v_lock);
if (bnp->bvn_vnp->v_count > 1) {
mutex_exit(&bnp->bvn_vnp->v_lock);
return (EBUSY);
}
mutex_exit(&bnp->bvn_vnp->v_lock);
}
kstat_delete(bfs->bfs_kstat);
bootfs_destruct(bfs);
list_destroy(&bfs->bfs_nodes);
kmem_free(bfs->bfs_mntpath, strlen(bfs->bfs_mntpath) + 1);
id_free(bootfs_idspace, bfs->bfs_minor);
kmem_free(bfs, sizeof (bootfs_t));
return (0);
}
static int
bootfs_root(vfs_t *vfsp, vnode_t **vpp)
{
bootfs_t *bfs;
bfs = (bootfs_t *)vfsp->vfs_data;
*vpp = bfs->bfs_rootvn->bvn_vnp;
VN_HOLD(*vpp)
return (0);
}
static int
bootfs_statvfs(vfs_t *vfsp, struct statvfs64 *sbp)
{
const bootfs_t *bfs = (bootfs_t *)vfsp;
dev32_t d32;
sbp->f_bsize = PAGESIZE;
sbp->f_frsize = PAGESIZE;
sbp->f_blocks = bfs->bfs_stat.bfss_nbytes.value.ui64 >> PAGESHIFT;
sbp->f_bfree = 0;
sbp->f_bavail = 0;
sbp->f_files = bfs->bfs_stat.bfss_nfiles.value.ui32 +
bfs->bfs_stat.bfss_ndirs.value.ui32;
sbp->f_ffree = 0;
sbp->f_favail = 0;
(void) cmpldev(&d32, vfsp->vfs_dev);
sbp->f_fsid = d32;
(void) strlcpy(sbp->f_basetype, bootfs_name, FSTYPSZ);
bzero(sbp->f_fstr, sizeof (sbp->f_fstr));
return (0);
}
static const fs_operation_def_t bootfs_vfsops_tmpl[] = {
VFSNAME_MOUNT, { .vfs_mount = bootfs_mount },
VFSNAME_UNMOUNT, { .vfs_unmount = bootfs_unmount },
VFSNAME_ROOT, { .vfs_root = bootfs_root },
VFSNAME_STATVFS, { .vfs_statvfs = bootfs_statvfs },
NULL, NULL
};
static int
bootfs_init(int fstype, char *name)
{
int ret;
bootfs_fstype = fstype;
ASSERT(bootfs_fstype != 0);
ret = vfs_setfsops(fstype, bootfs_vfsops_tmpl, NULL);
if (ret != 0)
return (ret);
ret = vn_make_ops(name, bootfs_vnodeops_template, &bootfs_vnodeops);
if (ret != 0) {
(void) vfs_freevfsops_by_type(bootfs_fstype);
return (ret);
}
bootfs_major = getudev();
if (bootfs_major == (major_t)-1) {
cmn_err(CE_WARN, "bootfs_init: Can't get unique device number");
bootfs_major = 0;
}
bootfs_nactive = 0;
return (0);
}
static mntopts_t bootfs_mntopts = {
0, NULL
};
static vfsdef_t bootfs_vfsdef = {
VFSDEF_VERSION,
"bootfs",
bootfs_init,
VSW_HASPROTO|VSW_STATS,
&bootfs_mntopts
};
static struct modlfs bootfs_modlfs = {
&mod_fsops, "boot-time modules file system", &bootfs_vfsdef
};
static struct modlinkage bootfs_modlinkage = {
MODREV_1, &bootfs_modlfs, NULL
};
int
_init(void)
{
bootfs_node_cache = kmem_cache_create("bootfs_node_cache",
sizeof (bootfs_node_t), 0, bootfs_node_constructor,
bootfs_node_destructor, NULL, NULL, NULL, 0);
bootfs_idspace = id_space_create("bootfs_minors", 1, INT32_MAX);
mutex_init(&bootfs_lock, NULL, MUTEX_DEFAULT, NULL);
return (mod_install(&bootfs_modlinkage));
}
int
_info(struct modinfo *modinfop)
{
return (mod_info(&bootfs_modlinkage, modinfop));
}
int
_fini(void)
{
int err;
mutex_enter(&bootfs_lock);
if (bootfs_nactive > 0) {
mutex_exit(&bootfs_lock);
return (EBUSY);
}
mutex_exit(&bootfs_lock);
err = mod_remove(&bootfs_modlinkage);
if (err != 0)
return (err);
(void) vfs_freevfsops_by_type(bootfs_fstype);
vn_freevnodeops(bootfs_vnodeops);
id_space_destroy(bootfs_idspace);
mutex_destroy(&bootfs_lock);
kmem_cache_destroy(bootfs_node_cache);
return (err);
}