vfs.c

来自「Linux Kernel 2.6.9 for OMAP1710」· C语言 代码 · 共 1,815 行 · 第 1/3 页

		 */
		v_mtime = verifier[0]&0x7fffffff;
		v_atime = verifier[1]&0x7fffffff;
		v_mode  = S_IFREG
			| ((verifier[0]&0x80000000) >> (32-7)) /* u+x */
			| ((verifier[1]&0x80000000) >> (32-9)) /* u+r */
			;
	}
	
	if (dchild->d_inode) {
		err = 0;

		switch (createmode) {
		case NFS3_CREATE_UNCHECKED:
			if (! S_ISREG(dchild->d_inode->i_mode))
				err = nfserr_exist;
			else if (truncp) {
				/* in nfsv4, we need to treat this case a little
				 * differently.  we don't want to truncate the
				 * file now; this would be wrong if the OPEN
				 * fails for some other reason.  furthermore,
				 * if the size is nonzero, we should ignore it
				 * according to spec!
				 */
				*truncp = (iap->ia_valid & ATTR_SIZE) && !iap->ia_size;
			}
			else {
				iap->ia_valid &= ATTR_SIZE;
				goto set_attr;
			}
			break;
		case NFS3_CREATE_EXCLUSIVE:
			if (   dchild->d_inode->i_mtime.tv_sec == v_mtime
			    && dchild->d_inode->i_atime.tv_sec == v_atime
			    && dchild->d_inode->i_mode  == v_mode
			    && dchild->d_inode->i_size  == 0 )
				break;
			 /* fallthru */
		case NFS3_CREATE_GUARDED:
			err = nfserr_exist;
		}
		goto out;
	}

	err = vfs_create(dirp, dchild, iap->ia_mode, NULL);
	if (err < 0)
		goto out_nfserr;

	if (EX_ISSYNC(fhp->fh_export)) {
		nfsd_sync_dir(dentry);
		/* setattr will sync the child (or not) */
	}

	/*
	 * Update the filehandle to get the new inode info.
	 */
	err = fh_update(resfhp);
	if (err)
		goto out;

	if (createmode == NFS3_CREATE_EXCLUSIVE) {
		/* Cram the verifier into atime/mtime/mode */
		iap->ia_valid = ATTR_MTIME|ATTR_ATIME
			| ATTR_MTIME_SET|ATTR_ATIME_SET
			| ATTR_MODE;
		/* XXX someone who knows this better please fix it for nsec */ 
		iap->ia_mtime.tv_sec = v_mtime;
		iap->ia_atime.tv_sec = v_atime;
		iap->ia_mtime.tv_nsec = 0;
		iap->ia_atime.tv_nsec = 0;
		iap->ia_mode  = v_mode;
	}

	/* Set file attributes.
	 * Mode has already been set but we might need to reset it
	 * for CREATE_EXCLUSIVE
	 * Irix appears to send along the gid when it tries to
	 * implement setgid directories via NFS. Clear out all that cruft.
	 */
 set_attr:
	if ((iap->ia_valid &= ~(ATTR_UID|ATTR_GID)) != 0)
 		err = nfsd_setattr(rqstp, resfhp, iap, 0, (time_t)0);

 out:
	fh_unlock(fhp);
	if (dchild && !IS_ERR(dchild))
		dput(dchild);
 	return err;
 
 out_nfserr:
	err = nfserrno(err);
	goto out;
}
#endif /* CONFIG_NFSD_V3 */

/*
 * Read a symlink. On entry, *lenp must contain the maximum path length that
 * fits into the buffer. On return, it contains the true length.
 * N.B. After this call fhp needs an fh_put
 */
int
nfsd_readlink(struct svc_rqst *rqstp, struct svc_fh *fhp, char *buf, int *lenp)
{
	struct dentry	*dentry;
	struct inode	*inode;
	mm_segment_t	oldfs;
	int		err;

	err = fh_verify(rqstp, fhp, S_IFLNK, MAY_NOP);
	if (err)
		goto out;

	dentry = fhp->fh_dentry;
	inode = dentry->d_inode;

	err = nfserr_inval;
	if (!inode->i_op || !inode->i_op->readlink)
		goto out;

	touch_atime(fhp->fh_export->ex_mnt, dentry);
	/* N.B. Why does this call need a get_fs()??
	 * Remove the set_fs and watch the fireworks:-) --okir
	 */

	oldfs = get_fs(); set_fs(KERNEL_DS);
	err = inode->i_op->readlink(dentry, buf, *lenp);
	set_fs(oldfs);

	if (err < 0)
		goto out_nfserr;
	*lenp = err;
	err = 0;
out:
	return err;

out_nfserr:
	err = nfserrno(err);
	goto out;
}

/*
 * Create a symlink and look up its inode
 * N.B. After this call _both_ fhp and resfhp need an fh_put
 */
int
nfsd_symlink(struct svc_rqst *rqstp, struct svc_fh *fhp,
				char *fname, int flen,
				char *path,  int plen,
				struct svc_fh *resfhp,
				struct iattr *iap)
{
	struct dentry	*dentry, *dnew;
	int		err, cerr;
	umode_t		mode;

	err = nfserr_noent;
	if (!flen || !plen)
		goto out;
	err = nfserr_exist;
	if (isdotent(fname, flen))
		goto out;

	err = fh_verify(rqstp, fhp, S_IFDIR, MAY_CREATE);
	if (err)
		goto out;
	fh_lock(fhp);
	dentry = fhp->fh_dentry;
	dnew = lookup_one_len(fname, dentry, flen);
	err = PTR_ERR(dnew);
	if (IS_ERR(dnew))
		goto out_nfserr;

	mode = S_IALLUGO;
	/* Only the MODE ATTRibute is even vaguely meaningful */
	if (iap && (iap->ia_valid & ATTR_MODE))
		mode = iap->ia_mode & S_IALLUGO;

	if (unlikely(path[plen] != 0)) {
		char *path_alloced = kmalloc(plen+1, GFP_KERNEL);
		if (path_alloced == NULL)
			err = -ENOMEM;
		else {
			strncpy(path_alloced, path, plen);
			path_alloced[plen] = 0;
			err = vfs_symlink(dentry->d_inode, dnew, path_alloced, mode);
			kfree(path_alloced);
		}
	} else
		err = vfs_symlink(dentry->d_inode, dnew, path, mode);

	if (!err) {
		if (EX_ISSYNC(fhp->fh_export))
			nfsd_sync_dir(dentry);
	} else
		err = nfserrno(err);
	fh_unlock(fhp);

	cerr = fh_compose(resfhp, fhp->fh_export, dnew, fhp);
	dput(dnew);
	if (err==0) err = cerr;
out:
	return err;

out_nfserr:
	err = nfserrno(err);
	goto out;
}

/*
 * Create a hardlink
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
int
nfsd_link(struct svc_rqst *rqstp, struct svc_fh *ffhp,
				char *name, int len, struct svc_fh *tfhp)
{
	struct dentry	*ddir, *dnew, *dold;
	struct inode	*dirp, *dest;
	int		err;

	err = fh_verify(rqstp, ffhp, S_IFDIR, MAY_CREATE);
	if (err)
		goto out;
	err = fh_verify(rqstp, tfhp, -S_IFDIR, MAY_NOP);
	if (err)
		goto out;

	err = nfserr_perm;
	if (!len)
		goto out;
	err = nfserr_exist;
	if (isdotent(name, len))
		goto out;

	fh_lock(ffhp);
	ddir = ffhp->fh_dentry;
	dirp = ddir->d_inode;

	dnew = lookup_one_len(name, ddir, len);
	err = PTR_ERR(dnew);
	if (IS_ERR(dnew))
		goto out_nfserr;

	dold = tfhp->fh_dentry;
	dest = dold->d_inode;

	err = vfs_link(dold, dirp, dnew);
	if (!err) {
		if (EX_ISSYNC(ffhp->fh_export)) {
			nfsd_sync_dir(ddir);
			write_inode_now(dest, 1);
		}
	} else {
		if (err == -EXDEV && rqstp->rq_vers == 2)
			err = nfserr_acces;
		else
			err = nfserrno(err);
	}

	fh_unlock(ffhp);
	dput(dnew);
out:
	return err;

out_nfserr:
	err = nfserrno(err);
	goto out;
}

/*
 * Rename a file
 * N.B. After this call _both_ ffhp and tfhp need an fh_put
 */
int
nfsd_rename(struct svc_rqst *rqstp, struct svc_fh *ffhp, char *fname, int flen,
			    struct svc_fh *tfhp, char *tname, int tlen)
{
	struct dentry	*fdentry, *tdentry, *odentry, *ndentry, *trap;
	struct inode	*fdir, *tdir;
	int		err;

	err = fh_verify(rqstp, ffhp, S_IFDIR, MAY_REMOVE);
	if (err)
		goto out;
	err = fh_verify(rqstp, tfhp, S_IFDIR, MAY_CREATE);
	if (err)
		goto out;

	fdentry = ffhp->fh_dentry;
	fdir = fdentry->d_inode;

	tdentry = tfhp->fh_dentry;
	tdir = tdentry->d_inode;

	err = (rqstp->rq_vers == 2) ? nfserr_acces : nfserr_xdev;
	if (fdir->i_sb != tdir->i_sb)
		goto out;

	err = nfserr_perm;
	if (!flen || isdotent(fname, flen) || !tlen || isdotent(tname, tlen))
		goto out;

	/* cannot use fh_lock as we need deadlock protective ordering
	 * so do it by hand */
	trap = lock_rename(tdentry, fdentry);
	ffhp->fh_locked = tfhp->fh_locked = 1;
	fill_pre_wcc(ffhp);
	fill_pre_wcc(tfhp);

	odentry = lookup_one_len(fname, fdentry, flen);
	err = PTR_ERR(odentry);
	if (IS_ERR(odentry))
		goto out_nfserr;

	err = -ENOENT;
	if (!odentry->d_inode)
		goto out_dput_old;
	err = -EINVAL;
	if (odentry == trap)
		goto out_dput_old;

	ndentry = lookup_one_len(tname, tdentry, tlen);
	err = PTR_ERR(ndentry);
	if (IS_ERR(ndentry))
		goto out_dput_old;
	err = -ENOTEMPTY;
	if (ndentry == trap)
		goto out_dput_new;

#ifdef MSNFS
	if ((ffhp->fh_export->ex_flags & NFSEXP_MSNFS) &&
		((atomic_read(&odentry->d_count) > 1)
		 || (atomic_read(&ndentry->d_count) > 1))) {
			err = nfserr_perm;
	} else
#endif
	err = vfs_rename(fdir, odentry, tdir, ndentry);
	if (!err && EX_ISSYNC(tfhp->fh_export)) {
		nfsd_sync_dir(tdentry);
		nfsd_sync_dir(fdentry);
	}

 out_dput_new:
	dput(ndentry);
 out_dput_old:
	dput(odentry);
 out_nfserr:
	if (err)
		err = nfserrno(err);

	/* we cannot reply on fh_unlock on the two filehandles,
	 * as that would do the wrong thing if the two directories
	 * were the same, so again we do it by hand
	 */
	fill_post_wcc(ffhp);
	fill_post_wcc(tfhp);
	unlock_rename(tdentry, fdentry);
	ffhp->fh_locked = tfhp->fh_locked = 0;

out:
	return err;
}

/*
 * Unlink a file or directory
 * N.B. After this call fhp needs an fh_put
 */
int
nfsd_unlink(struct svc_rqst *rqstp, struct svc_fh *fhp, int type,
				char *fname, int flen)
{
	struct dentry	*dentry, *rdentry;
	struct inode	*dirp;
	int		err;

	err = nfserr_acces;
	if (!flen || isdotent(fname, flen))
		goto out;
	err = fh_verify(rqstp, fhp, S_IFDIR, MAY_REMOVE);
	if (err)
		goto out;

	fh_lock(fhp);
	dentry = fhp->fh_dentry;
	dirp = dentry->d_inode;

	rdentry = lookup_one_len(fname, dentry, flen);
	err = PTR_ERR(rdentry);
	if (IS_ERR(rdentry))
		goto out_nfserr;

	if (!rdentry->d_inode) {
		dput(rdentry);
		err = nfserr_noent;
		goto out;
	}

	if (!type)
		type = rdentry->d_inode->i_mode & S_IFMT;

	if (type != S_IFDIR) { /* It's UNLINK */
#ifdef MSNFS
		if ((fhp->fh_export->ex_flags & NFSEXP_MSNFS) &&
			(atomic_read(&rdentry->d_count) > 1)) {
			err = nfserr_perm;
		} else
#endif
		err = vfs_unlink(dirp, rdentry);
	} else { /* It's RMDIR */
		err = vfs_rmdir(dirp, rdentry);
	}

	dput(rdentry);

	if (err)
		goto out_nfserr;
	if (EX_ISSYNC(fhp->fh_export)) 
		nfsd_sync_dir(dentry);

out:
	return err;

out_nfserr:
	err = nfserrno(err);
	goto out;
}

/*
 * Read entries from a directory.
 * The  NFSv3/4 verifier we ignore for now.
 */
int
nfsd_readdir(struct svc_rqst *rqstp, struct svc_fh *fhp, loff_t *offsetp, 
	     struct readdir_cd *cdp, encode_dent_fn func)
{
	int		err;
	struct file	*file;
	loff_t		offset = *offsetp;

	err = nfsd_open(rqstp, fhp, S_IFDIR, MAY_READ, &file);
	if (err)
		goto out;

	offset = vfs_llseek(file, offset, 0);
	if (offset < 0) {
		err = nfserrno((int)offset);
		goto out_close;
	}

	/*
	 * Read the directory entries. This silly loop is necessary because
	 * readdir() is not guaranteed to fill up the entire buffer, but
	 * may choose to do less.
	 */

	do {
		cdp->err = nfserr_eof; /* will be cleared on successful read */
		err = vfs_readdir(file, (filldir_t) func, cdp);
	} while (err >=0 && cdp->err == nfs_ok);
	if (err)
		err = nfserrno(err);
	else
		err = cdp->err;
	*offsetp = vfs_llseek(file, 0, 1);

	if (err == nfserr_eof || err == nfserr_toosmall)
		err = nfs_ok; /* can still be found in ->err */
out_close:
	nfsd_close(file);
out:
	return err;
}

/*
 * Get file system stats
 * N.B. After this call fhp needs an fh_put
 */
int
nfsd_statfs(struct svc_rqst *rqstp, struct svc_fh *fhp, struct kstatfs *stat)
{
	int err = fh_verify(rqstp, fhp, 0, MAY_NOP);
	if (!err && vfs_statfs(fhp->fh_dentry->d_inode->i_sb,stat))
		err = nfserr_io;
	return err;
}

/*
 * Check for a user's access permissions to this inode.
 */
int
nfsd_permission(struct svc_export *exp, struct dentry *dentry, int acc)
{
	struct inode	*inode = dentry->d_inode;
	int		err;

	if (acc == MAY_NOP)
		return 0;
#if 0
	dprintk("nfsd: permission 0x%x%s%s%s%s%s%s%s mode 0%o%s%s%s\n",
		acc,
		(acc & MAY_READ)?	" read"  : "",
		(acc & MAY_WRITE)?	" write" : "",
		(acc & MAY_EXEC)?	" exec"  : "",
		(acc & MAY_SATTR)?	" sattr" : "",
		(acc & MAY_TRUNC)?	" trunc" : "",
		(acc & MAY_LOCK)?	" lock"  : "",
		(acc & MAY_OWNER_OVERRIDE)? " owneroverride" : "",
		inode->i_mode,
		IS_IMMUTABLE(inode)?	" immut" : "",
		IS_APPEND(inode)?	" append" : "",
		IS_RDONLY(inode)?	" ro" : "");
	dprintk("      owner %d/%d user %d/%d\n",
		inode->i_uid, inode->i_gid, current->fsuid, current->fsgid);
#endif

	/* Normally we reject any write/sattr etc access on a read-only file
	 * system.  But if it is IRIX doing check on write-access for a 
	 * device special file, we ignore rofs.
	 */
	if (!(acc & MAY_LOCAL_ACCESS))
		if (acc & (MAY_WRITE | MAY_SATTR | MAY_TRUNC)) {
			if (EX_RDONLY(exp) || IS_RDONLY(inode))
				return nfserr_rofs;
			if (/* (acc & MAY_WRITE) && */ IS_IMMUTABLE(inode))
				return nfserr_perm;
		}
	if ((acc & MAY_TRUNC) && IS_APPEND(inode))
		return nfserr_perm;

	if (acc & MAY_LOCK) {
		/* If we cannot rely on authentication in NLM requests,
		 * just allow locks, otherwise require read permission, or
		 * ownership
		 */
		if (exp->ex_flags & NFSEXP_NOAUTHNLM)
			return 0;
		else
			acc = MAY_READ | MAY_OWNER_OVERRIDE;
	}
	/*
	 * The file owner always gets access permission for accesses that
	 * would normally be checked at open time. This is to make
	 * file access work even when the client has done a fchmod(fd, 0).
	 *
	 * However, `cp foo bar' should fail nevertheless when bar is
	 * readonly. A sensible way to do this might be to reject all
	 * attempts to truncate a read-only file, because a creat() call
	 * always implies file truncation.
	 * ... but this isn't really fair.  A process may reasonably call
	 * ftruncate on an open file descriptor on a file with perm 000.
	 * We must trust the client to do permission checking - using "ACCESS"
	 * with NFSv3.
	 */
	if ((acc & MAY_OWNER_OVERRIDE) &&
	    inode->i_uid == current->fsuid)
		return 0;

	err = permission(inode, acc & (MAY_READ|MAY_WRITE|MAY_EXEC), NULL);

	/* Allow read access to binaries even when mode 111 */
	if (err == -EACCES && S_ISREG(inode->i_mode) &&
	    acc == (MAY_READ | MAY_OWNER_OVERRIDE))
		err = permission(inode, MAY_EXEC, NULL);

	return err? nfserrno(err) : 0;
}

void
nfsd_racache_shutdown(void)
{
	if (!raparm_cache)
		return;
	dprintk("nfsd: freeing readahead buffers.\n");
	kfree(raparml);
	raparm_cache = raparml = NULL;
}
/*
 * Initialize readahead param cache
 */
int
nfsd_racache_init(int cache_size)
{
	int	i;

	if (raparm_cache)
		return 0;
	raparml = kmalloc(sizeof(struct raparms) * cache_size, GFP_KERNEL);

	if (raparml != NULL) {
		dprintk("nfsd: allocating %d readahead buffers.\n",
			cache_size);
		memset(raparml, 0, sizeof(struct raparms) * cache_size);
		for (i = 0; i < cache_size - 1; i++) {
			raparml[i].p_next = raparml + i + 1;
		}
		raparm_cache = raparml;
	} else {
		printk(KERN_WARNING
		       "nfsd: Could not allocate memory read-ahead cache.\n");
		return -ENOMEM;
	}
	nfsdstats.ra_size = cache_size;
	return 0;
}