minix/servers/mfs/path.c
David van Moolenbroek f76d75a5ec Various VFS and MFS fixes to improve correctness, consistency and
POSIX compliance.

VFS changes:
* truncate() on a file system mounted read-only no longer panics MFS.
* ftruncate() and fcntl(F_FREESP) now check for write permission on
  the file descriptor instead of the file, write().
* utime(), chown() and fchown() now check for file system read-only
  status.

MFS changes:
* link() and rename() no longer return the internal EENTERMOUNT and
  ELEAVEMOUNT errors to the application as part of a check on the
  source path.
* rename() now treats EENTERMOUNT from the destination path check as
  an error, preventing file system corruption from renaming a normal
  directory to an existing mountpoint directory.
* mountpoints (mounted-on dirs) are hidden better during lookups:
  - if a lookup starts from a mountpoint, the first component has to
    be ".." (anything else being a VFS-FS protocol violation).
  - in that case, the permissions of the mountpoint are not checked.
  - in all other cases, visiting a mountpoint always results in
    EENTERMOUNT.
* a lookup on ".." from a mount root or chroot(2) root no longer
  succeeds if the caller does not have search permission on that
  directory.
* POSIX: getdents() now updates directory access times.
* POSIX: readlink() now returns partial results instead of ERANGE.

Miscellaneous changes:
* semaphore file handling bug (leading to hangs) fixed in test 32.

The VFS changes should now put the burden of checking for read-only
status of file systems entirely on VFS, and limit the access
permission checks that file systems have to perform, to checking
search permission on directories during lookups. From this point on,
any deviation from that spceification should be considered a bug.
Note that for legacy reasons, the root partition is assumed to be
mounted read-write.
2009-05-18 11:27:12 +00:00

1407 lines
42 KiB
C

/* This file contains the procedures that look up path names in the directory
* system and determine the inode number that goes with a given path name.
*
* The entry points into this file are
* eat_path: the 'main' routine of the path-to-inode conversion mechanism
* last_dir: find the final directory on a given path
* advance: parse one component of a path name
* search_dir: search a directory for a string and return its inode number
*
*/
#include "fs.h"
#include <string.h>
#include <minix/callnr.h>
#include <minix/endpoint.h>
#include <sys/stat.h>
#include "buf.h"
#include "inode.h"
#include "super.h"
#include <minix/vfsif.h>
PUBLIC char dot1[2] = "."; /* used for search_dir to bypass the access */
PUBLIC char dot2[3] = ".."; /* permissions for . and .. */
FORWARD _PROTOTYPE( char *get_name, (char *old_name, char string [NAME_MAX]) );
FORWARD _PROTOTYPE( char *get_name_s, (char *name, char string[NAME_MAX+1]) );
FORWARD _PROTOTYPE( int ltraverse, (struct inode *rip, char *path,
char *suffix, int pathlen) );
FORWARD _PROTOTYPE( int ltraverse_s, (struct inode *rip, char *suffix) );
FORWARD _PROTOTYPE( int advance_s1, (struct inode *dirp,
char string[NAME_MAX], struct inode **resp) );
FORWARD _PROTOTYPE( int parse_path_s, (ino_t dir_ino, ino_t root_ino,
int flags, struct inode **res_inop,
size_t *offsetp, int *symlinkp) );
/*===========================================================================*
* lookup_o *
*===========================================================================*/
PUBLIC int lookup_o()
{
char string[PATH_MAX];
struct inode *rip;
int s_error, flags;
int len;
string[0] = '\0';
/* Check length. */
len = fs_m_in.REQ_PATH_LEN;
if(len > sizeof(user_path)) return E2BIG; /* too big for buffer */
if(len < 1) return EINVAL; /* too small for \0 */
/* Copy the pathname and set up caller's user and group id */
err_code = sys_datacopy(FS_PROC_NR, (vir_bytes) fs_m_in.REQ_PATH, SELF,
(vir_bytes) user_path, (phys_bytes) len);
if (err_code != OK) {
printf("mfs:%s:%d: sys_datacopy failed: %d\n", __FILE__, __LINE__, err_code);
return err_code;
}
/* Verify this is a null-terminated path. */
if(user_path[len-1] != '\0') {
printf("mfs:lookup: didn't get null-terminated string.\n");
return EINVAL;
}
caller_uid = fs_m_in.REQ_UID;
caller_gid = fs_m_in.REQ_GID;
flags = fs_m_in.REQ_FLAGS;
/* Clear RES_OFFSET for ENOENT */
fs_m_out.RES_OFFSET= 0;
/* Lookup inode */
rip = parse_path_o(user_path, string, flags);
/* Copy back the last name if it is required */
if (err_code != OK || (flags & PATH_PENULTIMATE)) {
s_error = sys_datacopy(SELF_E, (vir_bytes) string, FS_PROC_NR,
(vir_bytes) fs_m_in.REQ_USER_ADDR, (phys_bytes) NAME_MAX);
if (s_error != OK) {
printf("mfs:%s:%d: sys_datacopy failed: %d\n",
__FILE__, __LINE__, s_error);
return s_error;
}
}
/* Error or mount point encountered */
if (rip == NIL_INODE)
{
if (err_code != EENTERMOUNT)
fs_m_out.RES_INODE_NR = 0; /* signal no inode */
return err_code;
}
fs_m_out.RES_INODE_NR = rip->i_num;
fs_m_out.RES_MODE = rip->i_mode;
fs_m_out.RES_FILE_SIZE = rip->i_size;
if ( (rip->i_mode & I_TYPE) == I_BLOCK_SPECIAL) {
fs_m_out.RES_DEV = (dev_t) rip->i_zone[0];
}
/* Drop inode (path parse increased the counter) */
put_inode(rip);
return err_code;
}
/*===========================================================================*
* fs_lookup_s *
*===========================================================================*/
PUBLIC int fs_lookup_s()
{
cp_grant_id_t grant;
int r, r1, len, flags, symlinks;
size_t offset, size;
ino_t dir_ino, root_ino;
struct inode *rip;
grant= fs_m_in.REQ_L_GRANT;
size= fs_m_in.REQ_L_PATH_SIZE; /* Size of the buffer */
len = fs_m_in.REQ_L_PATH_LEN; /* including terminating nul */
offset= fs_m_in.REQ_L_PATH_OFF; /* offset in buffer */
dir_ino= fs_m_in.REQ_L_DIR_INO;
root_ino= fs_m_in.REQ_L_ROOT_INO;
flags = fs_m_in.REQ_L_FLAGS;
caller_uid = fs_m_in.REQ_L_UID;
caller_gid = fs_m_in.REQ_L_GID;
/* Check length. */
if(len > sizeof(user_path)) return E2BIG; /* too big for buffer */
if(len < 1)
{
printf("mfs:fs_lookup_s: string too small.\n");
return EINVAL; /* too small */
}
/* Copy the pathname and set up caller's user and group id */
r = sys_safecopyfrom(FS_PROC_NR, grant, offset,
(vir_bytes) user_path, (phys_bytes) len, D);
if (r != OK) {
printf("mfs:fs_lookup_s: sys_safecopyfrom failed: %d\n", r);
return r;
}
/* Verify this is a null-terminated path. */
if(user_path[len-1] != '\0') {
printf("mfs:fs_lookup_s: didn't get null-terminated string.\n");
return EINVAL;
}
#if 0
printf("mfs:fs_lookup_s: string '%s', ino %d, root %d\n",
user_path, dir_ino, root_ino);
#endif
/* Lookup inode */
rip= NULL;
r = parse_path_s(dir_ino, root_ino, flags, &rip, &offset, &symlinks);
if (symlinks != 0 && (r == ELEAVEMOUNT || r == EENTERMOUNT || r == ESYMLINK))
{
len= strlen(user_path)+1;
if (len > size)
return ENAMETOOLONG;
r1 = sys_safecopyto(FS_PROC_NR, grant, 0,
(vir_bytes) user_path, (phys_bytes) len, D);
if (r1 != OK) {
printf("mfs:fs_lookup_s: sys_safecopyto failed: %d\n", r1);
return r1;
}
#if 0
printf("mfs:fs_lookup_s: copied back path '%s', offset %d\n",
user_path, offset);
#endif
}
if (r == ELEAVEMOUNT || r == ESYMLINK)
{
/* Report offset and the error */
fs_m_out.RES_OFFSET = offset;
fs_m_out.RES_SYMLOOP = symlinks;
#if 0
printf("mfs:fs_lookup_s: returning %d, offset %d\n", r, offset);
#endif
if (rip) panic(__FILE__, "fs_lookup_s: rip should be clear",
(unsigned)rip);
return r;
}
if (r != OK && r != EENTERMOUNT)
{
if (rip) panic(__FILE__, "fs_lookup_s: rip should be clear",
(unsigned)rip);
return r;
}
fs_m_out.RES_INODE_NR = rip->i_num;
fs_m_out.RES_MODE = rip->i_mode;
fs_m_out.RES_FILE_SIZE = rip->i_size;
fs_m_out.RES_OFFSET = offset;
fs_m_out.RES_SYMLOOP2 = symlinks;
fs_m_out.RES_UID = rip->i_uid;
fs_m_out.RES_GID = rip->i_gid;
/* This is only valid for block and character specials. But it doesn't
* cause any harm to set RES_DEV always.
*/
fs_m_out.RES_DEV = (dev_t) rip->i_zone[0];
if (r == EENTERMOUNT)
put_inode(rip); /* Only return a reference to the final object */
return r;
}
/*===========================================================================*
* parse_path_o *
*===========================================================================*/
PUBLIC struct inode *parse_path_o(path, string, action)
char *path; /* the path name to be parsed */
char string[NAME_MAX]; /* the final component is returned here */
int action; /* action on last part of path */
{
/* This is the actual code for last_dir and eat_path. Return the inode of
* the last directory and the name of object within that directory, or the
* inode of the last object (an empty name will be returned). Names are
* returned in string. If string is null the name is discarded. The action
* code determines how "last" is defined. If an error occurs, NIL_INODE
* will be returned with an error code in err_code.
*/
struct inode *rip, *dir_ip;
struct inode *ver_rip;
char *new_name;
char lstring[NAME_MAX];
/* Find starting inode inode according to the request message */
if ((rip = find_inode(fs_dev, fs_m_in.REQ_INODE_NR)) == NIL_INODE) {
printf("mfs:parse_path: couldn't find starting inode %d for %s\n",
fs_m_in.REQ_INODE_NR, user_path);
err_code = ENOENT;
return NIL_INODE;
}
/* Find chroot inode according to the request message */
if (fs_m_in.REQ_CHROOT_NR != 0) {
if ((chroot_dir = find_inode(fs_dev, fs_m_in.REQ_CHROOT_NR))
== NIL_INODE) {
printf("FS: couldn't find chroot inode\n");
err_code = ENOENT;
return NIL_INODE;
}
}
else chroot_dir = NIL_INODE;
/* Set user and group ID */
caller_uid = fs_m_in.REQ_UID;
caller_gid = fs_m_in.REQ_GID;
/* No characters were processed yet */
path_processed = 0;
/* Current number of symlinks encountered */
Xsymloop = fs_m_in.REQ_SYMLOOP;
/* If dir has been removed return ENOENT. */
/* Note: empty (start) path is checked in the VFS process */
if (rip->i_nlinks == 0/* || *path == '\0'*/) {
err_code = ENOENT;
return(NIL_INODE);
}
/* There is only one way how the starting directory of the lookup
* can be a mount point which is not a root directory,
* namely: climbing up on a mount (ELEAVEMOUNT).
* In this case the lookup is intrested in the parent dir of the mount
* point, but the last ".." component was processed in the 'previous'
* FS process. Let's do that first.
*/
if (rip->i_mountpoint && rip->i_num != ROOT_INODE) {
dir_ip = rip;
rip = advance_o(&dir_ip, "..");
if (rip == NIL_INODE)
{
return NIL_INODE;
}
put_inode(rip); /* advance() increased the counter */
}
dup_inode(rip); /* inode will be returned with put_inode */
/* Looking for the starting directory?
* Note: this happens after EENTERMOUNT or ELEAVEMOUNT
* without more path component */
if (*path == '\0') {
return rip;
}
if (string == (char *) 0) string = lstring;
/* Scan the path component by component. */
while (TRUE) {
int slashes = 0;
/* Extract one component. Skip slashes first. */
while (path[slashes] == '/') {
slashes++;
path_processed++;
}
fs_m_out.RES_OFFSET = path_processed; /* For ENOENT */
if ( (new_name = get_name(path+slashes, string)) == (char*) 0) {
put_inode(rip); /* bad path in user space */
return(NIL_INODE);
}
if (*new_name == '\0' && (action & PATH_PENULTIMATE)) {
if ( (rip->i_mode & I_TYPE) == I_DIRECTORY) {
return(rip); /* normal exit */
} else {
/* last file of path prefix is not a directory */
put_inode(rip);
err_code = ENOTDIR;
return(NIL_INODE);
}
}
/* There is more path. Keep parsing. */
dir_ip = rip;
rip = advance_o(&dir_ip, string);
/* Mount point encountered? */
if (rip == NIL_INODE && (err_code == EENTERMOUNT ||
err_code == ELEAVEMOUNT)) {
put_inode(dir_ip);
return NIL_INODE;
}
if (rip == NIL_INODE) {
if (*new_name == '\0' && (action & PATH_NONSYMBOLIC) != 0)
{
return(dir_ip);
}
else if (err_code == ENOENT)
{
return(dir_ip);
}
else {
put_inode(dir_ip);
return(NIL_INODE);
}
}
/* The call to advance() succeeded. Fetch next component. */
if (S_ISLNK(rip->i_mode)) {
if (*new_name != '\0' || (action & PATH_OPAQUE) == 0) {
if (*new_name != '\0') new_name--;
/* Extract path name from the symlink file */
if (ltraverse(rip, user_path, new_name,
sizeof(user_path)) != OK) {
put_inode(dir_ip);
err_code = ENOENT;
return NIL_INODE;
}
/* Symloop limit reached? */
if (++Xsymloop > SYMLOOP_MAX) {
put_inode(dir_ip);
err_code = ELOOP;
return NIL_INODE;
}
/* Start over counting */
path_processed = 0;
/* Check whether new path is relative or absolute */
if (user_path[0] == '/') {
/* Go back to VFS */
put_inode(dir_ip);
err_code = ESYMLINK;
fs_m_out.RES_OFFSET = path_processed;
fs_m_out.RES_SYMLOOP = Xsymloop;
return NIL_INODE;
}
/* Path is relative */
else {
rip = dir_ip;
path = user_path;
continue;
}
}
}
else if (*new_name != '\0') {
put_inode(dir_ip);
path = new_name;
continue;
}
/* Either last name reached or symbolic link is opaque */
if ((action & PATH_NONSYMBOLIC) != 0) {
put_inode(rip);
return(dir_ip);
} else {
put_inode(dir_ip);
return(rip);
}
}
}
/*===========================================================================*
* parse_path_s *
*===========================================================================*/
PRIVATE int parse_path_s(dir_ino, root_ino, flags, res_inop, offsetp, symlinkp)
ino_t dir_ino;
ino_t root_ino;
int flags;
struct inode **res_inop;
size_t *offsetp;
int *symlinkp;
{
/* Parse the path in user_path, starting at dir_ino. If the path is the empty
* string, just return dir_ino. It is upto the caller to treat an empty
* path in a special way. Otherwise, if the path consists of just one or
* more slash ('/') characters, the path is replaced with ".". Otherwise,
* just look up the first (or only) component in path after skipping any
* leading slashes.
*/
int r, leaving_mount;
struct inode *rip, *dir_ip;
char *cp, *ncp;
char string[NAME_MAX+1];
#if 0
struct inode *ver_rip;
char *new_name;
#endif
/* Find starting inode inode according to the request message */
if ((rip = find_inode(fs_dev, dir_ino)) == NIL_INODE) {
printf("mfs:parse_path_s: couldn't find starting inode\n");
return ENOENT;
}
dup_inode(rip);
/* No characters were processed yet */
cp= user_path;
/* No symlinks encountered yet */
*symlinkp = 0;
/* If dir has been removed return ENOENT. */
if (rip->i_nlinks == 0) {
put_inode(rip);
return ENOENT;
}
/* If the given start inode is a mountpoint, we must be here because the file
* system mounted on top returned an ELEAVEMOUNT error. In this case, we must
* only accept ".." as the first path component.
*/
leaving_mount = rip->i_mountpoint;
/* Scan the path component by component. */
while (TRUE) {
if (cp[0] == '\0')
{
/* Empty path */
*res_inop= rip;
*offsetp += cp-user_path;
/* Return EENTERMOUNT if we are at a mount point */
if (rip->i_mountpoint)
{
return EENTERMOUNT;
}
return OK;
}
if (cp[0] == '/')
{
/* Special case code. If the remaining path consists of just
* slashes, we need to look up '.'
*/
while(cp[0] == '/')
cp++;
if (cp[0] == '\0')
{
strcpy(string, ".");
ncp= cp;
}
else
ncp= get_name_s(cp, string);
}
else
{
/* Just get the first component */
ncp= get_name_s(cp, string);
}
/* Special code for '..'. A process is not allowed to leave a chrooted
* environment. A lookup of '..' at the root of a mounted filesystem
* has to return ELEAVEMOUNT. In both cases, the caller needs search
* permission for the current inode, as it is used as directory.
*/
if (strcmp(string, "..") == 0)
{
if (rip->i_num == root_ino)
{
/* 'rip' is now accessed as directory */
if ((r = forbidden(rip, X_BIT)) != OK) {
put_inode(rip);
return r;
}
cp= ncp;
continue; /* Just ignore the '..' at a process'
* root.
*/
}
if (rip->i_num == ROOT_INODE && !rip->i_sp->s_is_root) {
/* Climbing up mountpoint */
/* 'rip' is now accessed as directory */
if ((r = forbidden(rip, X_BIT)) != OK) {
put_inode(rip);
return r;
}
put_inode(rip);
*offsetp += cp-user_path;
return ELEAVEMOUNT;
}
}
else
{
/* Check for misbehaving child file systems. */
if (leaving_mount) {
printf("mfs:parse_path_s: first component after "
"leaving mount is '%s'\n", string);
/* DO NOT pass back EENTERMOUNT. We supposedly got here
* because the child file system treated the last path
* component as "..". It is likely to do that again.
*/
put_inode(rip);
return EINVAL;
}
}
/* Only check for a mount point if we are not coming from one. */
if (!leaving_mount && rip->i_mountpoint)
{
*res_inop= rip;
*offsetp += cp-user_path;
return EENTERMOUNT;
}
/* There is more path. Keep parsing.
* If we're leaving a mountpoint, skip directory permission checks.
*/
dir_ip = rip;
r = advance_s1(dir_ip, leaving_mount ? dot2 : string, &rip);
if (r != OK)
{
put_inode(dir_ip);
return r;
}
leaving_mount = 0;
/* The call to advance() succeeded. Fetch next component. */
if (S_ISLNK(rip->i_mode)) {
if (ncp[0] == '\0' && (flags & PATH_RET_SYMLINK))
{
put_inode(dir_ip);
*res_inop= rip;
*offsetp += ncp-user_path;
return OK;
}
/* Extract path name from the symlink file */
r= ltraverse_s(rip, ncp);
ncp= user_path;
/* Symloop limit reached? */
if (++(*symlinkp) > SYMLOOP_MAX)
r= ELOOP;
/* Extract path name from the symlink file */
if (r != OK)
{
put_inode(dir_ip);
put_inode(rip);
return r;
}
if (ncp[0] == '/')
{
put_inode(dir_ip);
put_inode(rip);
*offsetp= 0;
return ESYMLINK;
}
put_inode(rip);
dup_inode(dir_ip);
rip= dir_ip;
}
put_inode(dir_ip);
cp= ncp;
}
}
/*===========================================================================*
* ltraverse *
*===========================================================================*/
PRIVATE int ltraverse(rip, path, suffix, pathlen)
register struct inode *rip; /* symbolic link */
char *path; /* path containing link */
char *suffix; /* suffix following link within path */
int pathlen;
{
/* Traverse a symbolic link. Copy the link text from the inode and insert
* the text into the path. Return error code or report success. Base
* directory has to be determined according to the first character of the
* new pathname.
*/
block_t b; /* block containing link text */
struct buf *bp; /* buffer containing link text */
size_t sl; /* length of link */
size_t tl; /* length of suffix */
char *sp; /* start of link text */
int r = OK;
bp = NIL_BUF;
if ((b = read_map(rip, (off_t) 0)) != NO_BLOCK) {
bp = get_block(rip->i_dev, b, NORMAL);
sl = rip->i_size;
sp = bp->b_data;
/* Insert symbolic text into path name. */
tl = strlen(suffix);
if (sl > 0 && sl + tl <= PATH_MAX-1) {
if(sl+tl >= pathlen)
panic(__FILE__,"path too small for symlink", sl+tl);
memmove(path+sl, suffix, tl);
memmove(path, sp, sl);
path[sl+tl] = 0;
/* Copy back to VFS layer THIS SHOULD BE IN parse_path.
* sys_datacopy() error, if any, gets returned as r later.
*/
r = sys_datacopy(SELF_E, (vir_bytes) path, FS_PROC_NR,
(vir_bytes) vfs_slink_storage, (phys_bytes) sl+tl+1);
/*
dup_inode(bip = path[0] == '/' ? chroot_dir : ldip);
*/
if(r != OK) {
printf("mfs:%s:%d: sys_datacopy failed: %d\n",
__FILE__, __LINE__, r);
}
} else panic(__FILE__,"didn't copy symlink", sl+tl);
}
else {
r = ENOENT;
}
put_block(bp, DIRECTORY_BLOCK);
put_inode(rip);
return r;
}
/*===========================================================================*
* ltraverse_s *
*===========================================================================*/
PRIVATE int ltraverse_s(rip, suffix)
register struct inode *rip; /* symbolic link */
char *suffix; /* current remaining path. Has to point in the
* user_path buffer
*/
{
/* Traverse a symbolic link. Copy the link text from the inode and insert
* the text into the path. Return error code or report success. Base
* directory has to be determined according to the first character of the
* new pathname.
*/
block_t b; /* block containing link text */
size_t sl; /* length of link */
size_t tl; /* length of suffix */
struct buf *bp; /* buffer containing link text */
char *sp; /* start of link text */
#if 0
int r = OK;
#endif
bp = NIL_BUF;
if ((b = read_map(rip, (off_t) 0)) == NO_BLOCK)
return EIO;
bp = get_block(rip->i_dev, b, NORMAL);
sl = rip->i_size;
sp = bp->b_data;
tl = strlen(suffix);
if (tl > 0)
{
/* For simplicity we require that suffix starts with a slash */
if (suffix[0] != '/')
{
panic(__FILE__,
"ltraverse_s: suffix does not start with a slash",
NO_NUM);
}
/* Move suffix to the right place */
if (sl + tl + 1 > sizeof(user_path))
return ENAMETOOLONG;
if (suffix-user_path != sl)
memmove(&user_path[sl], suffix, tl+1);
}
else
{
/* Set terminating nul */
user_path[sl]= '\0';
}
memmove(user_path, sp, sl);
#if 0
printf("mfs:ltraverse_s: new path '%s'\n", user_path);
#endif
put_block(bp, DIRECTORY_BLOCK);
return OK;
}
/*===========================================================================*
* advance_nocheck *
*===========================================================================*/
PUBLIC struct inode *advance_nocheck(pdirp, string)
struct inode **pdirp; /* inode for directory to be searched */
char string[NAME_MAX]; /* component name to look for */
{
/* Given a directory and a component of a path, look up the component in
* the directory, find the inode, open it, and return a pointer to its inode
* slot. If it can't be done, return NIL_INODE.
*/
register struct inode *rip, *dirp;
register struct super_block *sp;
int r, inumb;
dev_t mnt_dev;
ino_t numb;
dirp = *pdirp;
/* If 'string' is empty, yield same inode straight away. */
if (string[0] == '\0') { return(get_inode(dirp->i_dev, (int) dirp->i_num)); }
/* Check for NIL_INODE. */
if (dirp == NIL_INODE) { return(NIL_INODE); }
/* If 'string' is not present in the directory, signal error. */
if ( (r = search_dir_nocheck(dirp, string, &numb, LOOK_UP)) != OK) {
err_code = r;
return(NIL_INODE);
}
/* Don't go beyond the current root directory, unless the string is dot2.
* Note: it has to be checked only if this FS process owns the chroot
* directory of the process */
if (chroot_dir != NIL_INODE) {
if (dirp == chroot_dir && strcmp(string, "..") == 0 && string != dot2)
return(get_inode(dirp->i_dev, (int) dirp->i_num));
}
/* The component has been found in the directory. Get inode. */
if ( (rip = get_inode(dirp->i_dev, (int) numb)) == NIL_INODE) {
return(NIL_INODE);
}
/* The following test is for "mountpoint/.." where mountpoint is a
* mountpoint. ".." will refer to the root of the mounted filesystem,
* but has to become a reference to the parent of the 'mountpoint'
* directory.
*
* This case is recognized by the looked up name pointing to a
* root inode, and the directory in which it is held being a
* root inode, _and_ the name[1] being '.'. (This is a test for '..'
* and excludes '.'.)
*/
if (rip->i_num == ROOT_INODE) {
if (dirp->i_num == ROOT_INODE) {
if (string[1] == '.') {
sp = rip->i_sp;
if (!sp->s_is_root) {
/*printf("FSadvance: ELEAVEMOUNT callnr: %d, cp: %d, restp: %s\n",
call_nr, Xpath_processed, user_path + Xpath_processed);*/
/* Climbing up mountpoint */
err_code = ELEAVEMOUNT;
/* This will be the FS process endoint */
fs_m_out.m_source = rip->i_dev;
fs_m_out.RES_OFFSET = path_processed;
fs_m_out.RES_SYMLOOP = Xsymloop;
put_inode(rip);
/*put_inode(dirp);*/
rip = NIL_INODE;
}
}
}
}
if (rip == NIL_INODE) return(NIL_INODE);
/* See if the inode is mounted on. If so, switch to root directory of the
* mounted file system. The super_block provides the linkage between the
* inode mounted on and the root directory of the mounted file system.
*/
if (rip != NIL_INODE && rip->i_mountpoint) {
/* Mountpoint encountered, report it */
err_code = EENTERMOUNT;
fs_m_out.RES_INODE_NR = rip->i_num;
fs_m_out.RES_OFFSET = path_processed;
fs_m_out.RES_SYMLOOP = Xsymloop;
put_inode(rip);
rip = NIL_INODE;
}
return(rip); /* return pointer to inode's component */
}
/*===========================================================================*
* advance_o *
*===========================================================================*/
PUBLIC struct inode *advance_o(pdirp, string)
struct inode **pdirp; /* inode for directory to be searched */
char string[NAME_MAX]; /* component name to look for */
{
/* Given a directory and a component of a path, look up the component in
* the directory, find the inode, open it, and return a pointer to its inode
* slot. If it can't be done, return NIL_INODE.
*/
register struct inode *rip, *dirp;
register struct super_block *sp;
int r, inumb;
dev_t mnt_dev;
ino_t numb;
dirp = *pdirp;
/* If 'string' is empty, yield same inode straight away. */
if (string[0] == '\0') { return(get_inode(dirp->i_dev, (int) dirp->i_num)); }
/* Check for NIL_INODE. */
if (dirp == NIL_INODE) { return(NIL_INODE); }
/* If 'string' is not present in the directory, signal error. */
if ( (r = search_dir(dirp, string, &numb, LOOK_UP)) != OK) {
err_code = r;
return(NIL_INODE);
}
/* Don't go beyond the current root directory, unless the string is dot2.
* Note: it has to be checked only if this FS process owns the chroot
* directory of the process */
if (chroot_dir != NIL_INODE) {
if (dirp == chroot_dir && strcmp(string, "..") == 0 && string != dot2)
return(get_inode(dirp->i_dev, (int) dirp->i_num));
}
/* The component has been found in the directory. Get inode. */
if ( (rip = get_inode(dirp->i_dev, (int) numb)) == NIL_INODE) {
return(NIL_INODE);
}
/* The following test is for "mountpoint/.." where mountpoint is a
* mountpoint. ".." will refer to the root of the mounted filesystem,
* but has to become a reference to the parent of the 'mountpoint'
* directory.
*
* This case is recognized by the looked up name pointing to a
* root inode, and the directory in which it is held being a
* root inode, _and_ the name[1] being '.'. (This is a test for '..'
* and excludes '.'.)
*/
if (rip->i_num == ROOT_INODE) {
if (dirp->i_num == ROOT_INODE) {
if (string[1] == '.') {
sp = rip->i_sp;
if (!sp->s_is_root) {
/* Climbing up mountpoint */
err_code = ELEAVEMOUNT;
/* This will be the FS process endoint */
fs_m_out.m_source = rip->i_dev;
fs_m_out.RES_OFFSET = path_processed;
fs_m_out.RES_SYMLOOP = Xsymloop;
put_inode(rip);
/*put_inode(dirp);*/
rip = NIL_INODE;
}
}
}
}
if (rip == NIL_INODE) return(NIL_INODE);
/* See if the inode is mounted on. If so, switch to root directory of the
* mounted file system. The super_block provides the linkage between the
* inode mounted on and the root directory of the mounted file system.
*/
if (rip != NIL_INODE && rip->i_mountpoint) {
/* Mountpoint encountered, report it */
err_code = EENTERMOUNT;
fs_m_out.RES_INODE_NR = rip->i_num;
fs_m_out.RES_OFFSET = path_processed;
fs_m_out.RES_SYMLOOP = Xsymloop;
put_inode(rip);
rip = NIL_INODE;
}
return(rip); /* return pointer to inode's component */
}
/*===========================================================================*
* advance_s1 *
*===========================================================================*/
PRIVATE int advance_s1(dirp, string, resp)
struct inode *dirp; /* inode for directory to be searched */
char string[NAME_MAX]; /* component name to look for */
struct inode **resp; /* resulting inode */
{
/* Given a directory and a component of a path, look up the component in
* the directory, find the inode, open it, and return a pointer to its inode
* slot.
*/
int r;
ino_t numb;
struct inode *rip;
/* If 'string' is empty, return an error. */
if (string[0] == '\0') return ENOENT;
/* Check for NIL_INODE. */
if (dirp == NIL_INODE) panic(__FILE__, "advance_s: nil dirp", NO_NUM);
/* If 'string' is not present in the directory, signal error. */
if ( (r = search_dir(dirp, string, &numb, LOOK_UP)) != OK) {
return(r);
}
/* The component has been found in the directory. Get inode. */
if ( (rip = get_inode(dirp->i_dev, (int) numb)) == NIL_INODE) {
return(err_code);
}
*resp= rip;
return OK;
}
/*===========================================================================*
* get_name *
*===========================================================================*/
PRIVATE char *get_name(old_name, string)
char *old_name; /* path name to parse */
char string[NAME_MAX]; /* component extracted from 'old_name' */
{
/* Given a pointer to a path name in fs space, 'old_name', copy the next
* component to 'string' and pad with zeros. A pointer to that part of
* the name as yet unparsed is returned. Roughly speaking,
* 'get_name' = 'old_name' - 'string'.
*
* This routine follows the standard convention that /usr/ast, /usr//ast,
* //usr///ast and /usr/ast/ are all equivalent.
*/
register int c;
register char *np, *rnp;
np = string; /* 'np' points to current position */
rnp = old_name; /* 'rnp' points to unparsed string */
c = *rnp;
/* Copy the unparsed path, 'old_name', to the array, 'string'. */
while ( rnp < &old_name[PATH_MAX] && c != '/' && c != '\0') {
if (np < &string[NAME_MAX]) *np++ = c;
c = *++rnp; /* advance to next character */
path_processed++; /* count characters */
}
/* To make /usr/ast/ equivalent to /usr/ast, skip trailing slashes. */
while (c == '/' && rnp < &old_name[PATH_MAX]) {
c = *++rnp;
path_processed++; /* count characters */
}
if (np < &string[NAME_MAX]) *np = '\0'; /* Terminate string */
if (rnp >= &old_name[PATH_MAX]) {
err_code = ENAMETOOLONG;
return((char *) 0);
}
return(rnp);
}
/*===========================================================================*
* get_name_s *
*===========================================================================*/
PRIVATE char *get_name_s(path_name, string)
char *path_name; /* path name to parse */
char string[NAME_MAX+1]; /* component extracted from 'old_name' */
{
/* Given a pointer to a path name in fs space, 'path_name', copy the first
* component to 'string' (truncated if necessary, always nul terminated).
* A pointer to the string after the first component of the name as yet
* unparsed is returned. Roughly speaking,
* 'get_name_s' = 'path_name' - 'string'.
*
* This routine follows the standard convention that /usr/ast, /usr//ast,
* //usr///ast and /usr/ast/ are all equivalent.
*/
size_t len;
char *cp, *ep;
cp= path_name;
/* Skip leading slashes */
while (cp[0] == '/')
cp++;
/* Find the end of the first component */
ep= cp;
while(ep[0] != '\0' && ep[0] != '/')
ep++;
len= ep-cp;
/* Truncate the amount to be copied if it exceeds NAME_MAX */
if (len > NAME_MAX)
len= NAME_MAX;
/* Special case of the string at cp is empty */
if (len == 0)
{
/* Return "." */
strcpy(string, ".");
}
else
{
memcpy(string, cp, len);
string[len]= '\0';
}
return ep;
}
/*===========================================================================*
* search_dir *
*===========================================================================*/
PUBLIC int search_dir(ldir_ptr, string, numb, flag)
register struct inode *ldir_ptr; /* ptr to inode for dir to search */
char string[NAME_MAX]; /* component to search for */
ino_t *numb; /* pointer to inode number */
int flag; /* LOOK_UP, ENTER, DELETE or IS_EMPTY */
{
/* This function searches the directory whose inode is pointed to by 'ldip':
* if (flag == ENTER) enter 'string' in the directory with inode # '*numb';
* if (flag == DELETE) delete 'string' from the directory;
* if (flag == LOOK_UP) search for 'string' and return inode # in 'numb';
* if (flag == IS_EMPTY) return OK if only . and .. in dir else ENOTEMPTY;
*
* if 'string' is dot1 or dot2, no access permissions are checked.
*/
register struct direct *dp = NULL;
register struct buf *bp = NULL;
int i, r, e_hit, t, match;
mode_t bits;
off_t pos;
unsigned new_slots, old_slots;
block_t b;
struct super_block *sp;
int extended = 0;
/* If 'ldir_ptr' is not a pointer to a dir inode, error. */
if ( (ldir_ptr->i_mode & I_TYPE) != I_DIRECTORY) {
return(ENOTDIR);
}
r = OK;
if (flag != IS_EMPTY) {
bits = (flag == LOOK_UP ? X_BIT : W_BIT | X_BIT);
if (string == dot1 || string == dot2) {
if (flag != LOOK_UP) r = read_only(ldir_ptr);
/* only a writable device is required. */
}
else r = forbidden(ldir_ptr, bits); /* check access permissions */
}
if (r != OK) return(r);
/* Step through the directory one block at a time. */
old_slots = (unsigned) (ldir_ptr->i_size/DIR_ENTRY_SIZE);
new_slots = 0;
e_hit = FALSE;
match = 0; /* set when a string match occurs */
for (pos = 0; pos < ldir_ptr->i_size; pos += ldir_ptr->i_sp->s_block_size) {
b = read_map(ldir_ptr, pos); /* get block number */
/* Since directories don't have holes, 'b' cannot be NO_BLOCK. */
bp = get_block(ldir_ptr->i_dev, b, NORMAL); /* get a dir block */
if (bp == NO_BLOCK)
panic(__FILE__,"get_block returned NO_BLOCK", NO_NUM);
/* Search a directory block. */
for (dp = &bp->b_dir[0];
dp < &bp->b_dir[NR_DIR_ENTRIES(ldir_ptr->i_sp->s_block_size)];
dp++) {
if (++new_slots > old_slots) { /* not found, but room left */
if (flag == ENTER) e_hit = TRUE;
break;
}
/* Match occurs if string found. */
if (flag != ENTER && dp->d_ino != 0) {
if (flag == IS_EMPTY) {
/* If this test succeeds, dir is not empty. */
if (strcmp(dp->d_name, "." ) != 0 &&
strcmp(dp->d_name, "..") != 0) match = 1;
} else {
if (strncmp(dp->d_name, string, NAME_MAX) == 0){
match = 1;
}
}
}
if (match) {
/* LOOK_UP or DELETE found what it wanted. */
r = OK;
if (flag == IS_EMPTY) r = ENOTEMPTY;
else if (flag == DELETE) {
/* Save d_ino for recovery. */
t = NAME_MAX - sizeof(ino_t);
*((ino_t *) &dp->d_name[t]) = dp->d_ino;
dp->d_ino = 0; /* erase entry */
bp->b_dirt = DIRTY;
ldir_ptr->i_update |= CTIME | MTIME;
ldir_ptr->i_dirt = DIRTY;
} else {
sp = ldir_ptr->i_sp; /* 'flag' is LOOK_UP */
*numb = conv4(sp->s_native, (int) dp->d_ino);
}
put_block(bp, DIRECTORY_BLOCK);
return(r);
}
/* Check for free slot for the benefit of ENTER. */
if (flag == ENTER && dp->d_ino == 0) {
e_hit = TRUE; /* we found a free slot */
break;
}
}
/* The whole block has been searched or ENTER has a free slot. */
if (e_hit) break; /* e_hit set if ENTER can be performed now */
put_block(bp, DIRECTORY_BLOCK); /* otherwise, continue searching dir */
}
/* The whole directory has now been searched. */
if (flag != ENTER) {
return(flag == IS_EMPTY ? OK : ENOENT);
}
/* This call is for ENTER. If no free slot has been found so far, try to
* extend directory.
*/
if (e_hit == FALSE) { /* directory is full and no room left in last block */
new_slots++; /* increase directory size by 1 entry */
if (new_slots == 0) return(EFBIG); /* dir size limited by slot count */
if ( (bp = new_block(ldir_ptr, ldir_ptr->i_size)) == NIL_BUF)
return(err_code);
dp = &bp->b_dir[0];
extended = 1;
}
/* 'bp' now points to a directory block with space. 'dp' points to slot. */
(void) memset(dp->d_name, 0, (size_t) NAME_MAX); /* clear entry */
for (i = 0; i < NAME_MAX && string[i]; i++) dp->d_name[i] = string[i];
sp = ldir_ptr->i_sp;
dp->d_ino = conv4(sp->s_native, (int) *numb);
bp->b_dirt = DIRTY;
put_block(bp, DIRECTORY_BLOCK);
ldir_ptr->i_update |= CTIME | MTIME; /* mark mtime for update later */
ldir_ptr->i_dirt = DIRTY;
if (new_slots > old_slots) {
ldir_ptr->i_size = (off_t) new_slots * DIR_ENTRY_SIZE;
/* Send the change to disk if the directory is extended. */
if (extended) rw_inode(ldir_ptr, WRITING);
}
return(OK);
}
/*===========================================================================*
* search_dir_nocheck *
*===========================================================================*/
PUBLIC int search_dir_nocheck(ldir_ptr, string, numb, flag)
register struct inode *ldir_ptr; /* ptr to inode for dir to search */
char string[NAME_MAX]; /* component to search for */
ino_t *numb; /* pointer to inode number */
int flag; /* LOOK_UP, ENTER, DELETE or IS_EMPTY */
{
/* This function searches the directory whose inode is pointed to by 'ldip':
* if (flag == ENTER) enter 'string' in the directory with inode # '*numb';
* if (flag == DELETE) delete 'string' from the directory;
* if (flag == LOOK_UP) search for 'string' and return inode # in 'numb';
* if (flag == IS_EMPTY) return OK if only . and .. in dir else ENOTEMPTY;
*
* if 'string' is dot1 or dot2, no access permissions are checked.
*/
register struct direct *dp = NULL;
register struct buf *bp = NULL;
int i, r, e_hit, t, match;
mode_t bits;
off_t pos;
unsigned new_slots, old_slots;
block_t b;
struct super_block *sp;
int extended = 0;
/* If 'ldir_ptr' is not a pointer to a dir inode, error. */
if ( (ldir_ptr->i_mode & I_TYPE) != I_DIRECTORY) {
return(ENOTDIR);
}
r = OK;
if (flag != IS_EMPTY) {
bits = (flag == LOOK_UP ? X_BIT : W_BIT | X_BIT);
if (string == dot1 || string == dot2) {
if (flag != LOOK_UP) r = read_only(ldir_ptr);
/* only a writable device is required. */
}
}
if (r != OK) return(r);
/* Step through the directory one block at a time. */
old_slots = (unsigned) (ldir_ptr->i_size/DIR_ENTRY_SIZE);
new_slots = 0;
e_hit = FALSE;
match = 0; /* set when a string match occurs */
for (pos = 0; pos < ldir_ptr->i_size; pos += ldir_ptr->i_sp->s_block_size) {
b = read_map(ldir_ptr, pos); /* get block number */
/* Since directories don't have holes, 'b' cannot be NO_BLOCK. */
bp = get_block(ldir_ptr->i_dev, b, NORMAL); /* get a dir block */
if (bp == NO_BLOCK)
panic(__FILE__,"get_block returned NO_BLOCK", NO_NUM);
/* Search a directory block. */
for (dp = &bp->b_dir[0];
dp < &bp->b_dir[NR_DIR_ENTRIES(ldir_ptr->i_sp->s_block_size)];
dp++) {
if (++new_slots > old_slots) { /* not found, but room left */
if (flag == ENTER) e_hit = TRUE;
break;
}
/* Match occurs if string found. */
if (flag != ENTER && dp->d_ino != 0) {
if (flag == IS_EMPTY) {
/* If this test succeeds, dir is not empty. */
if (strcmp(dp->d_name, "." ) != 0 &&
strcmp(dp->d_name, "..") != 0) match = 1;
} else {
if (strncmp(dp->d_name, string, NAME_MAX) == 0){
match = 1;
}
}
}
if (match) {
/* LOOK_UP or DELETE found what it wanted. */
r = OK;
if (flag == IS_EMPTY) r = ENOTEMPTY;
else if (flag == DELETE) {
/* Save d_ino for recovery. */
t = NAME_MAX - sizeof(ino_t);
*((ino_t *) &dp->d_name[t]) = dp->d_ino;
dp->d_ino = 0; /* erase entry */
bp->b_dirt = DIRTY;
ldir_ptr->i_update |= CTIME | MTIME;
ldir_ptr->i_dirt = DIRTY;
} else {
sp = ldir_ptr->i_sp; /* 'flag' is LOOK_UP */
*numb = conv4(sp->s_native, (int) dp->d_ino);
}
put_block(bp, DIRECTORY_BLOCK);
return(r);
}
/* Check for free slot for the benefit of ENTER. */
if (flag == ENTER && dp->d_ino == 0) {
e_hit = TRUE; /* we found a free slot */
break;
}
}
/* The whole block has been searched or ENTER has a free slot. */
if (e_hit) break; /* e_hit set if ENTER can be performed now */
put_block(bp, DIRECTORY_BLOCK); /* otherwise, continue searching dir */
}
/* The whole directory has now been searched. */
if (flag != ENTER) {
return(flag == IS_EMPTY ? OK : ENOENT);
}
/* This call is for ENTER. If no free slot has been found so far, try to
* extend directory.
*/
if (e_hit == FALSE) { /* directory is full and no room left in last block */
new_slots++; /* increase directory size by 1 entry */
if (new_slots == 0) return(EFBIG); /* dir size limited by slot count */
if ( (bp = new_block(ldir_ptr, ldir_ptr->i_size)) == NIL_BUF)
return(err_code);
dp = &bp->b_dir[0];
extended = 1;
}
/* 'bp' now points to a directory block with space. 'dp' points to slot. */
(void) memset(dp->d_name, 0, (size_t) NAME_MAX); /* clear entry */
for (i = 0; i < NAME_MAX && string[i]; i++) dp->d_name[i] = string[i];
sp = ldir_ptr->i_sp;
dp->d_ino = conv4(sp->s_native, (int) *numb);
bp->b_dirt = DIRTY;
put_block(bp, DIRECTORY_BLOCK);
ldir_ptr->i_update |= CTIME | MTIME; /* mark mtime for update later */
ldir_ptr->i_dirt = DIRTY;
if (new_slots > old_slots) {
ldir_ptr->i_size = (off_t) new_slots * DIR_ENTRY_SIZE;
/* Send the change to disk if the directory is extended. */
if (extended) rw_inode(ldir_ptr, WRITING);
}
return(OK);
}
/*===========================================================================*
* eat_path_o *
*===========================================================================*/
PUBLIC struct inode *eat_path_o(path)
char *path; /* the path name to be parsed */
{
/* Parse the path 'path' and put its inode in the inode table. If not possible,
* return NIL_INODE as function value and an error code in 'err_code'.
*/
return parse_path_o(path, (char *) 0, EAT_PATH);
}
/*===========================================================================*
* last_dir_o *
*===========================================================================*/
PUBLIC struct inode *last_dir_o(path, string)
char *path; /* the path name to be parsed */
char string[NAME_MAX]; /* the final component is returned here */
{
/* Given a path, 'path', located in the fs address space, parse it as
* far as the last directory, fetch the inode for the last directory into
* the inode table, and return a pointer to the inode. In
* addition, return the final component of the path in 'string'.
* If the last directory can't be opened, return NIL_INODE and
* the reason for failure in 'err_code'.
*/
return parse_path_o(path, string, LAST_DIR);
}