2006-10-25 15:40:36 +02:00
|
|
|
/* This file contains the routines related to vnodes.
|
|
|
|
* The entry points are:
|
2012-02-13 16:28:04 +01:00
|
|
|
*
|
2006-10-25 15:40:36 +02:00
|
|
|
* get_vnode - increase counter and get details of an inode
|
|
|
|
* get_free_vnode - get a pointer to a free vnode obj
|
2012-02-13 16:28:04 +01:00
|
|
|
* find_vnode - find a vnode according to the FS endpoint and the inode num.
|
2006-10-25 15:40:36 +02:00
|
|
|
* dup_vnode - duplicate vnode (i.e. increase counter)
|
2012-02-13 16:28:04 +01:00
|
|
|
* put_vnode - drop vnode (i.e. decrease counter)
|
2006-10-25 15:40:36 +02:00
|
|
|
*/
|
- Introduce support for sticky bit.
- Revise VFS-FS protocol and update VFS/MFS/ISOFS accordingly.
- Clean up MFS by removing old, dead code (backwards compatibility is broken by
the new VFS-FS protocol, anyway) and rewrite other parts. Also, make sure all
functions have proper banners and prototypes.
- VFS should always provide a (syntactically) valid path to the FS; no need for
the FS to do sanity checks when leaving/entering mount points.
- Fix several bugs in MFS:
- Several path lookup bugs in MFS.
- A link can be too big for the path buffer.
- A mountpoint can become inaccessible when the creation of a new inode
fails, because the inode already exists and is a mountpoint.
- Introduce support for supplemental groups.
- Add test 46 to test supplemental group functionality (and removed obsolete
suppl. tests from test 2).
- Clean up VFS (not everything is done yet).
- ISOFS now opens device read-only. This makes the -r flag in the mount command
unnecessary (but will still report to be mounted read-write).
- Introduce PipeFS. PipeFS is a new FS that handles all anonymous and
named pipes. However, named pipes still reside on the (M)FS, as they are part
of the file system on disk. To make this work VFS now has a concept of
'mapped' inodes, which causes read, write, truncate and stat requests to be
redirected to the mapped FS, and all other requests to the original FS.
2009-12-20 21:27:14 +01:00
|
|
|
|
2006-10-25 15:40:36 +02:00
|
|
|
#include "fs.h"
|
|
|
|
#include "vnode.h"
|
|
|
|
#include "vmnt.h"
|
2007-01-05 17:36:55 +01:00
|
|
|
#include "file.h"
|
2006-10-25 15:40:36 +02:00
|
|
|
#include <minix/vfsif.h>
|
2012-02-13 16:28:04 +01:00
|
|
|
#include <assert.h>
|
2006-10-25 15:40:36 +02:00
|
|
|
|
2009-01-26 18:43:59 +01:00
|
|
|
/* Is vnode pointer reasonable? */
|
2011-02-16 19:58:30 +01:00
|
|
|
#if NDEBUG
|
|
|
|
#define SANEVP(v)
|
|
|
|
#define CHECKVN(v)
|
|
|
|
#define ASSERTVP(v)
|
|
|
|
#else
|
2009-01-26 18:43:59 +01:00
|
|
|
#define SANEVP(v) ((((v) >= &vnode[0] && (v) < &vnode[NR_VNODES])))
|
|
|
|
|
2010-07-02 14:41:19 +02:00
|
|
|
#define BADVP(v, f, l) printf("%s:%d: bad vp %p\n", f, l, v)
|
2009-01-26 18:43:59 +01:00
|
|
|
|
|
|
|
/* vp check that returns 0 for use in check_vrefs() */
|
|
|
|
#define CHECKVN(v) if(!SANEVP(v)) { \
|
|
|
|
BADVP(v, __FILE__, __LINE__); \
|
|
|
|
return 0; \
|
|
|
|
}
|
|
|
|
|
|
|
|
/* vp check that panics */
|
|
|
|
#define ASSERTVP(v) if(!SANEVP(v)) { \
|
2010-03-05 16:05:11 +01:00
|
|
|
BADVP(v, __FILE__, __LINE__); panic("bad vp"); }
|
2011-02-16 19:58:30 +01:00
|
|
|
#endif
|
2007-01-05 17:36:55 +01:00
|
|
|
|
2012-02-13 16:28:04 +01:00
|
|
|
#if LOCK_DEBUG
|
|
|
|
/*===========================================================================*
|
|
|
|
* check_vnode_locks_by_me *
|
|
|
|
*===========================================================================*/
|
2012-03-25 20:25:53 +02:00
|
|
|
void check_vnode_locks_by_me(struct fproc *rfp)
|
2012-02-13 16:28:04 +01:00
|
|
|
{
|
|
|
|
/* Check whether this thread still has locks held on vnodes */
|
|
|
|
struct vnode *vp;
|
|
|
|
|
|
|
|
for (vp = &vnode[0]; vp < &vnode[NR_VNODES]; vp++) {
|
|
|
|
if (tll_locked_by_me(&vp->v_lock)) {
|
make vfs & filesystems use failable copying
Change the kernel to add features to vircopy and safecopies so that
transparent copy fixing won't happen to avoid deadlocks, and such copies
fail with EFAULT.
Transparently making copying work from filesystems (as normally done by
the kernel & VM when copying fails because of missing/readonly memory)
is problematic as it can happen that, for file-mapped ranges, that that
same filesystem that is blocked on the copy request is needed to satisfy
the memory range, leading to deadlock. Dito for VFS itself, if done with
a blocking call.
This change makes the copying done from a filesystem fail in such cases
with EFAULT by VFS adding the CPF_TRY flag to the grants. If a FS call
fails with EFAULT, VFS will then request the range to be made available
to VM after the FS is unblocked, allowing it to be used to satisfy the
range if need be in another VFS thread.
Similarly, for datacopies that VFS itself does, it uses the failable
vircopy variant and callers use a wrapper that talk to VM if necessary
to get the copy to work.
. kernel: add CPF_TRY flag to safecopies
. kernel: only request writable ranges to VM for the
target buffer when copying fails
. do copying in VFS TRY-first
. some fixes in VM to build SANITYCHECK mode
. add regression test for the cases where
- a FS system call needs memory mapped in a process that the
FS itself must map.
- such a range covers more than one file-mapped region.
. add 'try' mode to vircopy, physcopy
. add flags field to copy kernel call messages
. if CP_FLAG_TRY is set, do not transparently try
to fix memory ranges
. for use by VFS when accessing user buffers to avoid
deadlock
. remove some obsolete backwards compatability assignments
. VFS: let thread scheduling work for VM requests too
Allows VFS to make calls to VM while suspending and resuming
the currently running thread. Does currently not work for the
main thread.
. VM: add fix memory range call for use by VFS
Change-Id: I295794269cea51a3163519a9cfe5901301d90b32
2014-01-16 14:22:13 +01:00
|
|
|
panic("Thread %d still holds vnode lock on vp %p call_nr=%d\n",
|
2012-04-13 14:50:38 +02:00
|
|
|
mthread_self(), vp, job_call_nr);
|
2012-02-13 16:28:04 +01:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (rfp->fp_vp_rdlocks != 0)
|
|
|
|
panic("Thread %d still holds read locks on a vnode (%d) call_nr=%d\n",
|
2012-04-13 14:50:38 +02:00
|
|
|
mthread_self(), rfp->fp_vp_rdlocks, job_call_nr);
|
2012-02-13 16:28:04 +01:00
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
/*===========================================================================*
|
|
|
|
* check_vnode_locks *
|
|
|
|
*===========================================================================*/
|
2012-03-25 20:25:53 +02:00
|
|
|
void check_vnode_locks()
|
2012-02-13 16:28:04 +01:00
|
|
|
{
|
|
|
|
struct vnode *vp;
|
|
|
|
int count = 0;
|
|
|
|
|
|
|
|
for (vp = &vnode[0]; vp < &vnode[NR_VNODES]; vp++)
|
|
|
|
if (is_vnode_locked(vp)) {
|
|
|
|
count++;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (count) panic("%d locked vnodes\n", count);
|
|
|
|
#if 0
|
|
|
|
printf("check_vnode_locks OK\n");
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
2006-10-25 15:40:36 +02:00
|
|
|
/*===========================================================================*
|
|
|
|
* get_free_vnode *
|
|
|
|
*===========================================================================*/
|
2012-03-25 20:25:53 +02:00
|
|
|
struct vnode *get_free_vnode()
|
2006-10-25 15:40:36 +02:00
|
|
|
{
|
- Introduce support for sticky bit.
- Revise VFS-FS protocol and update VFS/MFS/ISOFS accordingly.
- Clean up MFS by removing old, dead code (backwards compatibility is broken by
the new VFS-FS protocol, anyway) and rewrite other parts. Also, make sure all
functions have proper banners and prototypes.
- VFS should always provide a (syntactically) valid path to the FS; no need for
the FS to do sanity checks when leaving/entering mount points.
- Fix several bugs in MFS:
- Several path lookup bugs in MFS.
- A link can be too big for the path buffer.
- A mountpoint can become inaccessible when the creation of a new inode
fails, because the inode already exists and is a mountpoint.
- Introduce support for supplemental groups.
- Add test 46 to test supplemental group functionality (and removed obsolete
suppl. tests from test 2).
- Clean up VFS (not everything is done yet).
- ISOFS now opens device read-only. This makes the -r flag in the mount command
unnecessary (but will still report to be mounted read-write).
- Introduce PipeFS. PipeFS is a new FS that handles all anonymous and
named pipes. However, named pipes still reside on the (M)FS, as they are part
of the file system on disk. To make this work VFS now has a concept of
'mapped' inodes, which causes read, write, truncate and stat requests to be
redirected to the mapped FS, and all other requests to the original FS.
2009-12-20 21:27:14 +01:00
|
|
|
/* Find a free vnode slot in the vnode table (it's not actually allocated) */
|
2006-10-25 15:40:36 +02:00
|
|
|
struct vnode *vp;
|
|
|
|
|
- Introduce support for sticky bit.
- Revise VFS-FS protocol and update VFS/MFS/ISOFS accordingly.
- Clean up MFS by removing old, dead code (backwards compatibility is broken by
the new VFS-FS protocol, anyway) and rewrite other parts. Also, make sure all
functions have proper banners and prototypes.
- VFS should always provide a (syntactically) valid path to the FS; no need for
the FS to do sanity checks when leaving/entering mount points.
- Fix several bugs in MFS:
- Several path lookup bugs in MFS.
- A link can be too big for the path buffer.
- A mountpoint can become inaccessible when the creation of a new inode
fails, because the inode already exists and is a mountpoint.
- Introduce support for supplemental groups.
- Add test 46 to test supplemental group functionality (and removed obsolete
suppl. tests from test 2).
- Clean up VFS (not everything is done yet).
- ISOFS now opens device read-only. This makes the -r flag in the mount command
unnecessary (but will still report to be mounted read-write).
- Introduce PipeFS. PipeFS is a new FS that handles all anonymous and
named pipes. However, named pipes still reside on the (M)FS, as they are part
of the file system on disk. To make this work VFS now has a concept of
'mapped' inodes, which causes read, write, truncate and stat requests to be
redirected to the mapped FS, and all other requests to the original FS.
2009-12-20 21:27:14 +01:00
|
|
|
for (vp = &vnode[0]; vp < &vnode[NR_VNODES]; ++vp) {
|
2012-02-13 16:28:04 +01:00
|
|
|
if (vp->v_ref_count == 0 && !is_vnode_locked(vp)) {
|
- Introduce support for sticky bit.
- Revise VFS-FS protocol and update VFS/MFS/ISOFS accordingly.
- Clean up MFS by removing old, dead code (backwards compatibility is broken by
the new VFS-FS protocol, anyway) and rewrite other parts. Also, make sure all
functions have proper banners and prototypes.
- VFS should always provide a (syntactically) valid path to the FS; no need for
the FS to do sanity checks when leaving/entering mount points.
- Fix several bugs in MFS:
- Several path lookup bugs in MFS.
- A link can be too big for the path buffer.
- A mountpoint can become inaccessible when the creation of a new inode
fails, because the inode already exists and is a mountpoint.
- Introduce support for supplemental groups.
- Add test 46 to test supplemental group functionality (and removed obsolete
suppl. tests from test 2).
- Clean up VFS (not everything is done yet).
- ISOFS now opens device read-only. This makes the -r flag in the mount command
unnecessary (but will still report to be mounted read-write).
- Introduce PipeFS. PipeFS is a new FS that handles all anonymous and
named pipes. However, named pipes still reside on the (M)FS, as they are part
of the file system on disk. To make this work VFS now has a concept of
'mapped' inodes, which causes read, write, truncate and stat requests to be
redirected to the mapped FS, and all other requests to the original FS.
2009-12-20 21:27:14 +01:00
|
|
|
vp->v_uid = -1;
|
|
|
|
vp->v_gid = -1;
|
|
|
|
vp->v_sdev = NO_DEV;
|
2012-02-13 16:28:04 +01:00
|
|
|
vp->v_mapfs_e = NONE;
|
|
|
|
vp->v_mapfs_count = 0;
|
- Introduce support for sticky bit.
- Revise VFS-FS protocol and update VFS/MFS/ISOFS accordingly.
- Clean up MFS by removing old, dead code (backwards compatibility is broken by
the new VFS-FS protocol, anyway) and rewrite other parts. Also, make sure all
functions have proper banners and prototypes.
- VFS should always provide a (syntactically) valid path to the FS; no need for
the FS to do sanity checks when leaving/entering mount points.
- Fix several bugs in MFS:
- Several path lookup bugs in MFS.
- A link can be too big for the path buffer.
- A mountpoint can become inaccessible when the creation of a new inode
fails, because the inode already exists and is a mountpoint.
- Introduce support for supplemental groups.
- Add test 46 to test supplemental group functionality (and removed obsolete
suppl. tests from test 2).
- Clean up VFS (not everything is done yet).
- ISOFS now opens device read-only. This makes the -r flag in the mount command
unnecessary (but will still report to be mounted read-write).
- Introduce PipeFS. PipeFS is a new FS that handles all anonymous and
named pipes. However, named pipes still reside on the (M)FS, as they are part
of the file system on disk. To make this work VFS now has a concept of
'mapped' inodes, which causes read, write, truncate and stat requests to be
redirected to the mapped FS, and all other requests to the original FS.
2009-12-20 21:27:14 +01:00
|
|
|
vp->v_mapinode_nr = 0;
|
|
|
|
return(vp);
|
2007-01-05 17:36:55 +01:00
|
|
|
}
|
2012-02-13 16:28:04 +01:00
|
|
|
}
|
2007-01-05 17:36:55 +01:00
|
|
|
|
2006-10-25 15:40:36 +02:00
|
|
|
err_code = ENFILE;
|
2010-05-10 15:26:00 +02:00
|
|
|
return(NULL);
|
2006-10-25 15:40:36 +02:00
|
|
|
}
|
|
|
|
|
- Introduce support for sticky bit.
- Revise VFS-FS protocol and update VFS/MFS/ISOFS accordingly.
- Clean up MFS by removing old, dead code (backwards compatibility is broken by
the new VFS-FS protocol, anyway) and rewrite other parts. Also, make sure all
functions have proper banners and prototypes.
- VFS should always provide a (syntactically) valid path to the FS; no need for
the FS to do sanity checks when leaving/entering mount points.
- Fix several bugs in MFS:
- Several path lookup bugs in MFS.
- A link can be too big for the path buffer.
- A mountpoint can become inaccessible when the creation of a new inode
fails, because the inode already exists and is a mountpoint.
- Introduce support for supplemental groups.
- Add test 46 to test supplemental group functionality (and removed obsolete
suppl. tests from test 2).
- Clean up VFS (not everything is done yet).
- ISOFS now opens device read-only. This makes the -r flag in the mount command
unnecessary (but will still report to be mounted read-write).
- Introduce PipeFS. PipeFS is a new FS that handles all anonymous and
named pipes. However, named pipes still reside on the (M)FS, as they are part
of the file system on disk. To make this work VFS now has a concept of
'mapped' inodes, which causes read, write, truncate and stat requests to be
redirected to the mapped FS, and all other requests to the original FS.
2009-12-20 21:27:14 +01:00
|
|
|
|
2006-10-25 15:40:36 +02:00
|
|
|
/*===========================================================================*
|
|
|
|
* find_vnode *
|
|
|
|
*===========================================================================*/
|
2012-04-02 17:16:44 +02:00
|
|
|
struct vnode *find_vnode(int fs_e, ino_t ino)
|
2006-10-25 15:40:36 +02:00
|
|
|
{
|
|
|
|
/* Find a specified (FS endpoint and inode number) vnode in the
|
|
|
|
* vnode table */
|
|
|
|
struct vnode *vp;
|
|
|
|
|
|
|
|
for (vp = &vnode[0]; vp < &vnode[NR_VNODES]; ++vp)
|
2012-02-13 16:28:04 +01:00
|
|
|
if (vp->v_ref_count > 0 && vp->v_inode_nr == ino && vp->v_fs_e == fs_e)
|
- Introduce support for sticky bit.
- Revise VFS-FS protocol and update VFS/MFS/ISOFS accordingly.
- Clean up MFS by removing old, dead code (backwards compatibility is broken by
the new VFS-FS protocol, anyway) and rewrite other parts. Also, make sure all
functions have proper banners and prototypes.
- VFS should always provide a (syntactically) valid path to the FS; no need for
the FS to do sanity checks when leaving/entering mount points.
- Fix several bugs in MFS:
- Several path lookup bugs in MFS.
- A link can be too big for the path buffer.
- A mountpoint can become inaccessible when the creation of a new inode
fails, because the inode already exists and is a mountpoint.
- Introduce support for supplemental groups.
- Add test 46 to test supplemental group functionality (and removed obsolete
suppl. tests from test 2).
- Clean up VFS (not everything is done yet).
- ISOFS now opens device read-only. This makes the -r flag in the mount command
unnecessary (but will still report to be mounted read-write).
- Introduce PipeFS. PipeFS is a new FS that handles all anonymous and
named pipes. However, named pipes still reside on the (M)FS, as they are part
of the file system on disk. To make this work VFS now has a concept of
'mapped' inodes, which causes read, write, truncate and stat requests to be
redirected to the mapped FS, and all other requests to the original FS.
2009-12-20 21:27:14 +01:00
|
|
|
return(vp);
|
2012-02-13 16:28:04 +01:00
|
|
|
|
2010-05-10 15:26:00 +02:00
|
|
|
return(NULL);
|
2006-10-25 15:40:36 +02:00
|
|
|
}
|
|
|
|
|
2012-02-13 16:28:04 +01:00
|
|
|
/*===========================================================================*
|
|
|
|
* is_vnode_locked *
|
|
|
|
*===========================================================================*/
|
2012-03-25 20:25:53 +02:00
|
|
|
int is_vnode_locked(struct vnode *vp)
|
2012-02-13 16:28:04 +01:00
|
|
|
{
|
|
|
|
/* Find out whether a thread holds a lock on this vnode or is trying to obtain
|
|
|
|
* a lock. */
|
|
|
|
ASSERTVP(vp);
|
|
|
|
|
|
|
|
return(tll_islocked(&vp->v_lock) || tll_haspendinglock(&vp->v_lock));
|
|
|
|
}
|
|
|
|
|
|
|
|
/*===========================================================================*
|
|
|
|
* init_vnodes *
|
|
|
|
*===========================================================================*/
|
2012-03-25 20:25:53 +02:00
|
|
|
void init_vnodes(void)
|
2012-02-13 16:28:04 +01:00
|
|
|
{
|
|
|
|
struct vnode *vp;
|
|
|
|
|
|
|
|
for (vp = &vnode[0]; vp < &vnode[NR_VNODES]; ++vp) {
|
|
|
|
vp->v_fs_e = NONE;
|
|
|
|
vp->v_mapfs_e = NONE;
|
|
|
|
vp->v_inode_nr = 0;
|
|
|
|
vp->v_ref_count = 0;
|
|
|
|
vp->v_fs_count = 0;
|
|
|
|
vp->v_mapfs_count = 0;
|
|
|
|
tll_init(&vp->v_lock);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*===========================================================================*
|
|
|
|
* lock_vnode *
|
|
|
|
*===========================================================================*/
|
2012-03-25 20:25:53 +02:00
|
|
|
int lock_vnode(struct vnode *vp, tll_access_t locktype)
|
2012-02-13 16:28:04 +01:00
|
|
|
{
|
|
|
|
int r;
|
|
|
|
|
|
|
|
ASSERTVP(vp);
|
|
|
|
|
|
|
|
r = tll_lock(&vp->v_lock, locktype);
|
|
|
|
|
|
|
|
#if LOCK_DEBUG
|
|
|
|
if (locktype == VNODE_READ) {
|
|
|
|
fp->fp_vp_rdlocks++;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
if (r == EBUSY) return(r);
|
|
|
|
return(OK);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*===========================================================================*
|
|
|
|
* unlock_vnode *
|
|
|
|
*===========================================================================*/
|
2012-03-25 20:25:53 +02:00
|
|
|
void unlock_vnode(struct vnode *vp)
|
2012-02-13 16:28:04 +01:00
|
|
|
{
|
|
|
|
#if LOCK_DEBUG
|
|
|
|
int i;
|
|
|
|
register struct vnode *rvp;
|
|
|
|
struct worker_thread *w;
|
|
|
|
#endif
|
|
|
|
ASSERTVP(vp);
|
|
|
|
|
|
|
|
#if LOCK_DEBUG
|
|
|
|
/* Decrease read-only lock counter when not locked as VNODE_OPCL or
|
|
|
|
* VNODE_WRITE */
|
|
|
|
if (!tll_locked_by_me(&vp->v_lock)) {
|
|
|
|
fp->fp_vp_rdlocks--;
|
|
|
|
}
|
|
|
|
|
|
|
|
for (i = 0; i < NR_VNODES; i++) {
|
|
|
|
rvp = &vnode[i];
|
|
|
|
|
|
|
|
w = rvp->v_lock.t_write;
|
|
|
|
assert(w != self);
|
|
|
|
while (w && w->w_next != NULL) {
|
|
|
|
w = w->w_next;
|
|
|
|
assert(w != self);
|
|
|
|
}
|
|
|
|
|
|
|
|
w = rvp->v_lock.t_serial;
|
|
|
|
assert(w != self);
|
|
|
|
while (w && w->w_next != NULL) {
|
|
|
|
w = w->w_next;
|
|
|
|
assert(w != self);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
tll_unlock(&vp->v_lock);
|
|
|
|
}
|
2006-10-25 15:40:36 +02:00
|
|
|
|
VFS: fix locking bugs
.sync and fsync used unnecessarily restrictive locking type
.fsync violated locking order by obtaining a vmnt lock after a filp lock
.fsync contained a TOCTOU bug
.new_node violated locking rules (didn't upgrade lock upon file creation)
.do_pipe used unnecessarily restrictive locking type
.always lock pipes exclusively; even a read operation might require to do
a write on a vnode object (update pipe size)
.when opening a file with O_TRUNC, upgrade vnode lock when truncating
.utime used unnecessarily restrictive locking type
.path parsing:
.always acquire VMNT_WRITE or VMNT_EXCL on vmnt and downgrade to
VMNT_READ if that was what was actually requested. This prevents the
following deadlock scenario:
thread A:
lock_vmnt(vmp, TLL_READSER);
lock_vnode(vp, TLL_READSER);
upgrade_vmnt_lock(vmp, TLL_WRITE);
thread B:
lock_vmnt(vmp, TLL_READ);
lock_vnode(vp, TLL_READSER);
thread A will be stuck in upgrade_vmnt_lock and thread B is stuck in
lock_vnode. This happens when, for example, thread A tries create a
new node (open.c:new_node) and thread B tries to do eat_path to
change dir (stadir.c:do_chdir). When the path is being resolved, a
vnode is always locked with VNODE_OPCL (TLL_READSER) and then
downgraded to VNODE_READ if read-only is actually requested. Thread
A locks the vmnt with VMNT_WRITE (TLL_READSER) which still allows
VMNT_READ locks. Thread B can't acquire a lock on the vnode because
thread A has it; Thread A can't upgrade its vmnt lock to VMNT_WRITE
(TLL_WRITE) because thread B has a VMNT_READ lock on it.
By serializing vmnt locks during path parsing, thread B can only
acquire a lock on vmp when thread A has completely finished its
operation.
2012-11-30 13:49:53 +01:00
|
|
|
/*===========================================================================*
|
|
|
|
* vnode *
|
|
|
|
*===========================================================================*/
|
|
|
|
void upgrade_vnode_lock(struct vnode *vp)
|
|
|
|
{
|
|
|
|
ASSERTVP(vp);
|
|
|
|
tll_upgrade(&vp->v_lock);
|
|
|
|
}
|
|
|
|
|
2006-10-25 15:40:36 +02:00
|
|
|
/*===========================================================================*
|
|
|
|
* dup_vnode *
|
|
|
|
*===========================================================================*/
|
2012-03-25 20:25:53 +02:00
|
|
|
void dup_vnode(struct vnode *vp)
|
2006-10-25 15:40:36 +02:00
|
|
|
{
|
|
|
|
/* dup_vnode() is called to increment the vnode and therefore the
|
|
|
|
* referred inode's counter.
|
|
|
|
*/
|
2009-01-26 18:43:59 +01:00
|
|
|
ASSERTVP(vp);
|
2007-01-05 17:36:55 +01:00
|
|
|
vp->v_ref_count++;
|
2006-10-25 15:40:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/*===========================================================================*
|
|
|
|
* put_vnode *
|
|
|
|
*===========================================================================*/
|
2012-03-25 20:25:53 +02:00
|
|
|
void put_vnode(struct vnode *vp)
|
2006-10-25 15:40:36 +02:00
|
|
|
{
|
2012-02-13 16:28:04 +01:00
|
|
|
/* Decrease vnode's usage counter and decrease inode's usage counter in the
|
- Introduce support for sticky bit.
- Revise VFS-FS protocol and update VFS/MFS/ISOFS accordingly.
- Clean up MFS by removing old, dead code (backwards compatibility is broken by
the new VFS-FS protocol, anyway) and rewrite other parts. Also, make sure all
functions have proper banners and prototypes.
- VFS should always provide a (syntactically) valid path to the FS; no need for
the FS to do sanity checks when leaving/entering mount points.
- Fix several bugs in MFS:
- Several path lookup bugs in MFS.
- A link can be too big for the path buffer.
- A mountpoint can become inaccessible when the creation of a new inode
fails, because the inode already exists and is a mountpoint.
- Introduce support for supplemental groups.
- Add test 46 to test supplemental group functionality (and removed obsolete
suppl. tests from test 2).
- Clean up VFS (not everything is done yet).
- ISOFS now opens device read-only. This makes the -r flag in the mount command
unnecessary (but will still report to be mounted read-write).
- Introduce PipeFS. PipeFS is a new FS that handles all anonymous and
named pipes. However, named pipes still reside on the (M)FS, as they are part
of the file system on disk. To make this work VFS now has a concept of
'mapped' inodes, which causes read, write, truncate and stat requests to be
redirected to the mapped FS, and all other requests to the original FS.
2009-12-20 21:27:14 +01:00
|
|
|
* corresponding FS process. Decreasing the fs_count each time we decrease the
|
|
|
|
* ref count would lead to poor performance. Instead, only decrease fs_count
|
|
|
|
* when the ref count hits zero. However, this could lead to fs_count to wrap.
|
|
|
|
* To prevent this, we drop the counter to 1 when the counter hits 256.
|
|
|
|
* We maintain fs_count as a sanity check to make sure VFS and the FS are in
|
|
|
|
* sync.
|
2006-10-25 15:40:36 +02:00
|
|
|
*/
|
2012-02-13 16:28:04 +01:00
|
|
|
int r, lock_vp;
|
|
|
|
|
2009-01-26 18:43:59 +01:00
|
|
|
ASSERTVP(vp);
|
2006-10-25 15:40:36 +02:00
|
|
|
|
2012-02-13 16:28:04 +01:00
|
|
|
/* Lock vnode. It's quite possible this thread already has a lock on this
|
|
|
|
* vnode. That's no problem, because the reference counter will not decrease
|
|
|
|
* to zero in that case. However, if the counter does decrease to zero *and*
|
|
|
|
* is already locked, we have a consistency problem somewhere. */
|
|
|
|
lock_vp = lock_vnode(vp, VNODE_OPCL);
|
|
|
|
|
- Introduce support for sticky bit.
- Revise VFS-FS protocol and update VFS/MFS/ISOFS accordingly.
- Clean up MFS by removing old, dead code (backwards compatibility is broken by
the new VFS-FS protocol, anyway) and rewrite other parts. Also, make sure all
functions have proper banners and prototypes.
- VFS should always provide a (syntactically) valid path to the FS; no need for
the FS to do sanity checks when leaving/entering mount points.
- Fix several bugs in MFS:
- Several path lookup bugs in MFS.
- A link can be too big for the path buffer.
- A mountpoint can become inaccessible when the creation of a new inode
fails, because the inode already exists and is a mountpoint.
- Introduce support for supplemental groups.
- Add test 46 to test supplemental group functionality (and removed obsolete
suppl. tests from test 2).
- Clean up VFS (not everything is done yet).
- ISOFS now opens device read-only. This makes the -r flag in the mount command
unnecessary (but will still report to be mounted read-write).
- Introduce PipeFS. PipeFS is a new FS that handles all anonymous and
named pipes. However, named pipes still reside on the (M)FS, as they are part
of the file system on disk. To make this work VFS now has a concept of
'mapped' inodes, which causes read, write, truncate and stat requests to be
redirected to the mapped FS, and all other requests to the original FS.
2009-12-20 21:27:14 +01:00
|
|
|
if (vp->v_ref_count > 1) {
|
2007-01-05 17:36:55 +01:00
|
|
|
/* Decrease counter */
|
|
|
|
vp->v_ref_count--;
|
2012-02-13 16:28:04 +01:00
|
|
|
if (vp->v_fs_count > 256)
|
2007-01-05 17:36:55 +01:00
|
|
|
vnode_clean_refs(vp);
|
2012-02-13 16:28:04 +01:00
|
|
|
if (lock_vp != EBUSY) unlock_vnode(vp);
|
2007-01-05 17:36:55 +01:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2012-02-13 16:28:04 +01:00
|
|
|
/* If we already had a lock, there is a consistency problem */
|
|
|
|
assert(lock_vp != EBUSY);
|
VFS: fix locking bugs
.sync and fsync used unnecessarily restrictive locking type
.fsync violated locking order by obtaining a vmnt lock after a filp lock
.fsync contained a TOCTOU bug
.new_node violated locking rules (didn't upgrade lock upon file creation)
.do_pipe used unnecessarily restrictive locking type
.always lock pipes exclusively; even a read operation might require to do
a write on a vnode object (update pipe size)
.when opening a file with O_TRUNC, upgrade vnode lock when truncating
.utime used unnecessarily restrictive locking type
.path parsing:
.always acquire VMNT_WRITE or VMNT_EXCL on vmnt and downgrade to
VMNT_READ if that was what was actually requested. This prevents the
following deadlock scenario:
thread A:
lock_vmnt(vmp, TLL_READSER);
lock_vnode(vp, TLL_READSER);
upgrade_vmnt_lock(vmp, TLL_WRITE);
thread B:
lock_vmnt(vmp, TLL_READ);
lock_vnode(vp, TLL_READSER);
thread A will be stuck in upgrade_vmnt_lock and thread B is stuck in
lock_vnode. This happens when, for example, thread A tries create a
new node (open.c:new_node) and thread B tries to do eat_path to
change dir (stadir.c:do_chdir). When the path is being resolved, a
vnode is always locked with VNODE_OPCL (TLL_READSER) and then
downgraded to VNODE_READ if read-only is actually requested. Thread
A locks the vmnt with VMNT_WRITE (TLL_READSER) which still allows
VMNT_READ locks. Thread B can't acquire a lock on the vnode because
thread A has it; Thread A can't upgrade its vmnt lock to VMNT_WRITE
(TLL_WRITE) because thread B has a VMNT_READ lock on it.
By serializing vmnt locks during path parsing, thread B can only
acquire a lock on vmp when thread A has completely finished its
operation.
2012-11-30 13:49:53 +01:00
|
|
|
upgrade_vnode_lock(vp); /* Acquire exclusive access */
|
2012-02-13 16:28:04 +01:00
|
|
|
|
|
|
|
/* A vnode that's not in use can't be put back. */
|
|
|
|
if (vp->v_ref_count <= 0)
|
|
|
|
panic("put_vnode failed: bad v_ref_count %d\n", vp->v_ref_count);
|
- Introduce support for sticky bit.
- Revise VFS-FS protocol and update VFS/MFS/ISOFS accordingly.
- Clean up MFS by removing old, dead code (backwards compatibility is broken by
the new VFS-FS protocol, anyway) and rewrite other parts. Also, make sure all
functions have proper banners and prototypes.
- VFS should always provide a (syntactically) valid path to the FS; no need for
the FS to do sanity checks when leaving/entering mount points.
- Fix several bugs in MFS:
- Several path lookup bugs in MFS.
- A link can be too big for the path buffer.
- A mountpoint can become inaccessible when the creation of a new inode
fails, because the inode already exists and is a mountpoint.
- Introduce support for supplemental groups.
- Add test 46 to test supplemental group functionality (and removed obsolete
suppl. tests from test 2).
- Clean up VFS (not everything is done yet).
- ISOFS now opens device read-only. This makes the -r flag in the mount command
unnecessary (but will still report to be mounted read-write).
- Introduce PipeFS. PipeFS is a new FS that handles all anonymous and
named pipes. However, named pipes still reside on the (M)FS, as they are part
of the file system on disk. To make this work VFS now has a concept of
'mapped' inodes, which causes read, write, truncate and stat requests to be
redirected to the mapped FS, and all other requests to the original FS.
2009-12-20 21:27:14 +01:00
|
|
|
|
|
|
|
/* fs_count should indicate that the file is in use. */
|
2012-02-13 16:28:04 +01:00
|
|
|
if (vp->v_fs_count <= 0)
|
|
|
|
panic("put_vnode failed: bad v_fs_count %d\n", vp->v_fs_count);
|
2007-01-05 17:36:55 +01:00
|
|
|
|
- Introduce support for sticky bit.
- Revise VFS-FS protocol and update VFS/MFS/ISOFS accordingly.
- Clean up MFS by removing old, dead code (backwards compatibility is broken by
the new VFS-FS protocol, anyway) and rewrite other parts. Also, make sure all
functions have proper banners and prototypes.
- VFS should always provide a (syntactically) valid path to the FS; no need for
the FS to do sanity checks when leaving/entering mount points.
- Fix several bugs in MFS:
- Several path lookup bugs in MFS.
- A link can be too big for the path buffer.
- A mountpoint can become inaccessible when the creation of a new inode
fails, because the inode already exists and is a mountpoint.
- Introduce support for supplemental groups.
- Add test 46 to test supplemental group functionality (and removed obsolete
suppl. tests from test 2).
- Clean up VFS (not everything is done yet).
- ISOFS now opens device read-only. This makes the -r flag in the mount command
unnecessary (but will still report to be mounted read-write).
- Introduce PipeFS. PipeFS is a new FS that handles all anonymous and
named pipes. However, named pipes still reside on the (M)FS, as they are part
of the file system on disk. To make this work VFS now has a concept of
'mapped' inodes, which causes read, write, truncate and stat requests to be
redirected to the mapped FS, and all other requests to the original FS.
2009-12-20 21:27:14 +01:00
|
|
|
/* Tell FS we don't need this inode to be open anymore. */
|
2012-02-13 16:28:04 +01:00
|
|
|
r = req_putnode(vp->v_fs_e, vp->v_inode_nr, vp->v_fs_count);
|
2007-01-05 17:36:55 +01:00
|
|
|
|
2012-02-13 16:28:04 +01:00
|
|
|
if (r != OK) {
|
|
|
|
printf("VFS: putnode failed: %d\n", r);
|
|
|
|
util_stacktrace();
|
- Introduce support for sticky bit.
- Revise VFS-FS protocol and update VFS/MFS/ISOFS accordingly.
- Clean up MFS by removing old, dead code (backwards compatibility is broken by
the new VFS-FS protocol, anyway) and rewrite other parts. Also, make sure all
functions have proper banners and prototypes.
- VFS should always provide a (syntactically) valid path to the FS; no need for
the FS to do sanity checks when leaving/entering mount points.
- Fix several bugs in MFS:
- Several path lookup bugs in MFS.
- A link can be too big for the path buffer.
- A mountpoint can become inaccessible when the creation of a new inode
fails, because the inode already exists and is a mountpoint.
- Introduce support for supplemental groups.
- Add test 46 to test supplemental group functionality (and removed obsolete
suppl. tests from test 2).
- Clean up VFS (not everything is done yet).
- ISOFS now opens device read-only. This makes the -r flag in the mount command
unnecessary (but will still report to be mounted read-write).
- Introduce PipeFS. PipeFS is a new FS that handles all anonymous and
named pipes. However, named pipes still reside on the (M)FS, as they are part
of the file system on disk. To make this work VFS now has a concept of
'mapped' inodes, which causes read, write, truncate and stat requests to be
redirected to the mapped FS, and all other requests to the original FS.
2009-12-20 21:27:14 +01:00
|
|
|
}
|
|
|
|
|
2012-02-13 16:28:04 +01:00
|
|
|
/* This inode could've been mapped. If so, tell mapped FS to close it as
|
|
|
|
* well. If mapped onto same FS, this putnode is not needed. */
|
|
|
|
if (vp->v_mapfs_e != NONE && vp->v_mapfs_e != vp->v_fs_e)
|
|
|
|
req_putnode(vp->v_mapfs_e, vp->v_mapinode_nr, vp->v_mapfs_count);
|
|
|
|
|
- Introduce support for sticky bit.
- Revise VFS-FS protocol and update VFS/MFS/ISOFS accordingly.
- Clean up MFS by removing old, dead code (backwards compatibility is broken by
the new VFS-FS protocol, anyway) and rewrite other parts. Also, make sure all
functions have proper banners and prototypes.
- VFS should always provide a (syntactically) valid path to the FS; no need for
the FS to do sanity checks when leaving/entering mount points.
- Fix several bugs in MFS:
- Several path lookup bugs in MFS.
- A link can be too big for the path buffer.
- A mountpoint can become inaccessible when the creation of a new inode
fails, because the inode already exists and is a mountpoint.
- Introduce support for supplemental groups.
- Add test 46 to test supplemental group functionality (and removed obsolete
suppl. tests from test 2).
- Clean up VFS (not everything is done yet).
- ISOFS now opens device read-only. This makes the -r flag in the mount command
unnecessary (but will still report to be mounted read-write).
- Introduce PipeFS. PipeFS is a new FS that handles all anonymous and
named pipes. However, named pipes still reside on the (M)FS, as they are part
of the file system on disk. To make this work VFS now has a concept of
'mapped' inodes, which causes read, write, truncate and stat requests to be
redirected to the mapped FS, and all other requests to the original FS.
2009-12-20 21:27:14 +01:00
|
|
|
vp->v_fs_count = 0;
|
|
|
|
vp->v_ref_count = 0;
|
2009-12-21 10:36:34 +01:00
|
|
|
vp->v_mapfs_count = 0;
|
2012-02-13 16:28:04 +01:00
|
|
|
|
|
|
|
unlock_vnode(vp);
|
2006-10-25 15:40:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2007-01-05 17:36:55 +01:00
|
|
|
/*===========================================================================*
|
|
|
|
* vnode_clean_refs *
|
|
|
|
*===========================================================================*/
|
2012-03-25 20:25:53 +02:00
|
|
|
void vnode_clean_refs(struct vnode *vp)
|
2007-01-05 17:36:55 +01:00
|
|
|
{
|
|
|
|
/* Tell the underlying FS to drop all reference but one. */
|
|
|
|
|
2010-05-10 15:26:00 +02:00
|
|
|
if (vp == NULL) return;
|
- Introduce support for sticky bit.
- Revise VFS-FS protocol and update VFS/MFS/ISOFS accordingly.
- Clean up MFS by removing old, dead code (backwards compatibility is broken by
the new VFS-FS protocol, anyway) and rewrite other parts. Also, make sure all
functions have proper banners and prototypes.
- VFS should always provide a (syntactically) valid path to the FS; no need for
the FS to do sanity checks when leaving/entering mount points.
- Fix several bugs in MFS:
- Several path lookup bugs in MFS.
- A link can be too big for the path buffer.
- A mountpoint can become inaccessible when the creation of a new inode
fails, because the inode already exists and is a mountpoint.
- Introduce support for supplemental groups.
- Add test 46 to test supplemental group functionality (and removed obsolete
suppl. tests from test 2).
- Clean up VFS (not everything is done yet).
- ISOFS now opens device read-only. This makes the -r flag in the mount command
unnecessary (but will still report to be mounted read-write).
- Introduce PipeFS. PipeFS is a new FS that handles all anonymous and
named pipes. However, named pipes still reside on the (M)FS, as they are part
of the file system on disk. To make this work VFS now has a concept of
'mapped' inodes, which causes read, write, truncate and stat requests to be
redirected to the mapped FS, and all other requests to the original FS.
2009-12-20 21:27:14 +01:00
|
|
|
if (vp->v_fs_count <= 1) return; /* Nothing to do */
|
2007-01-05 17:36:55 +01:00
|
|
|
|
- Introduce support for sticky bit.
- Revise VFS-FS protocol and update VFS/MFS/ISOFS accordingly.
- Clean up MFS by removing old, dead code (backwards compatibility is broken by
the new VFS-FS protocol, anyway) and rewrite other parts. Also, make sure all
functions have proper banners and prototypes.
- VFS should always provide a (syntactically) valid path to the FS; no need for
the FS to do sanity checks when leaving/entering mount points.
- Fix several bugs in MFS:
- Several path lookup bugs in MFS.
- A link can be too big for the path buffer.
- A mountpoint can become inaccessible when the creation of a new inode
fails, because the inode already exists and is a mountpoint.
- Introduce support for supplemental groups.
- Add test 46 to test supplemental group functionality (and removed obsolete
suppl. tests from test 2).
- Clean up VFS (not everything is done yet).
- ISOFS now opens device read-only. This makes the -r flag in the mount command
unnecessary (but will still report to be mounted read-write).
- Introduce PipeFS. PipeFS is a new FS that handles all anonymous and
named pipes. However, named pipes still reside on the (M)FS, as they are part
of the file system on disk. To make this work VFS now has a concept of
'mapped' inodes, which causes read, write, truncate and stat requests to be
redirected to the mapped FS, and all other requests to the original FS.
2009-12-20 21:27:14 +01:00
|
|
|
/* Drop all references except one */
|
|
|
|
req_putnode(vp->v_fs_e, vp->v_inode_nr, vp->v_fs_count - 1);
|
|
|
|
vp->v_fs_count = 1;
|
2007-01-05 17:36:55 +01:00
|
|
|
}
|
- Introduce support for sticky bit.
- Revise VFS-FS protocol and update VFS/MFS/ISOFS accordingly.
- Clean up MFS by removing old, dead code (backwards compatibility is broken by
the new VFS-FS protocol, anyway) and rewrite other parts. Also, make sure all
functions have proper banners and prototypes.
- VFS should always provide a (syntactically) valid path to the FS; no need for
the FS to do sanity checks when leaving/entering mount points.
- Fix several bugs in MFS:
- Several path lookup bugs in MFS.
- A link can be too big for the path buffer.
- A mountpoint can become inaccessible when the creation of a new inode
fails, because the inode already exists and is a mountpoint.
- Introduce support for supplemental groups.
- Add test 46 to test supplemental group functionality (and removed obsolete
suppl. tests from test 2).
- Clean up VFS (not everything is done yet).
- ISOFS now opens device read-only. This makes the -r flag in the mount command
unnecessary (but will still report to be mounted read-write).
- Introduce PipeFS. PipeFS is a new FS that handles all anonymous and
named pipes. However, named pipes still reside on the (M)FS, as they are part
of the file system on disk. To make this work VFS now has a concept of
'mapped' inodes, which causes read, write, truncate and stat requests to be
redirected to the mapped FS, and all other requests to the original FS.
2009-12-20 21:27:14 +01:00
|
|
|
|