new helper: iterate_supers()
[safe/jmp/linux-2.6] / fs / gfs2 / inode.c
index 5a31d42..51d8061 100644 (file)
 #include "acl.h"
 #include "bmap.h"
 #include "dir.h"
-#include "eattr.h"
+#include "xattr.h"
 #include "glock.h"
 #include "glops.h"
 #include "inode.h"
 #include "log.h"
 #include "meta_io.h"
-#include "ops_address.h"
 #include "quota.h"
 #include "rgrp.h"
 #include "trans.h"
@@ -46,7 +45,7 @@ static int iget_test(struct inode *inode, void *opaque)
        struct gfs2_inode *ip = GFS2_I(inode);
        u64 *no_addr = opaque;
 
-       if (ip->i_no_addr == *no_addr && test_bit(GIF_USER, &ip->i_flags))
+       if (ip->i_no_addr == *no_addr)
                return 1;
 
        return 0;
@@ -59,7 +58,6 @@ static int iget_set(struct inode *inode, void *opaque)
 
        inode->i_ino = (unsigned long)*no_addr;
        ip->i_no_addr = *no_addr;
-       set_bit(GIF_USER, &ip->i_flags);
        return 0;
 }
 
@@ -85,7 +83,7 @@ static int iget_skip_test(struct inode *inode, void *opaque)
        struct gfs2_inode *ip = GFS2_I(inode);
        struct gfs2_skip_data *data = opaque;
 
-       if (ip->i_no_addr == data->no_addr && test_bit(GIF_USER, &ip->i_flags)){
+       if (ip->i_no_addr == data->no_addr{
                if (inode->i_state & (I_FREEING|I_CLEAR|I_WILL_FREE)){
                        data->skipped = 1;
                        return 0;
@@ -104,7 +102,6 @@ static int iget_skip_set(struct inode *inode, void *opaque)
                return 1;
        inode->i_ino = (unsigned long)(data->no_addr);
        ip->i_no_addr = data->no_addr;
-       set_bit(GIF_USER, &ip->i_flags);
        return 0;
 }
 
@@ -126,7 +123,7 @@ static struct inode *gfs2_iget_skip(struct super_block *sb,
  * directory entry when gfs2_inode_lookup() is invoked. Part of the code
  * segment inside gfs2_inode_lookup code needs to get moved around.
  *
- * Clean up I_LOCK and I_NEW as well.
+ * Clears I_NEW as well.
  **/
 
 void gfs2_set_iop(struct inode *inode)
@@ -161,7 +158,6 @@ void gfs2_set_iop(struct inode *inode)
  * @sb: The super block
  * @no_addr: The inode number
  * @type: The type of the inode
- * @skip_freeing: set this not return an inode if it is currently being freed.
  *
  * Returns: A VFS inode, or an error
  */
@@ -169,17 +165,14 @@ void gfs2_set_iop(struct inode *inode)
 struct inode *gfs2_inode_lookup(struct super_block *sb,
                                unsigned int type,
                                u64 no_addr,
-                               u64 no_formal_ino, int skip_freeing)
+                               u64 no_formal_ino)
 {
        struct inode *inode;
        struct gfs2_inode *ip;
        struct gfs2_glock *io_gl;
        int error;
 
-       if (skip_freeing)
-               inode = gfs2_iget_skip(sb, no_addr);
-       else
-               inode = gfs2_iget(sb, no_addr);
+       inode = gfs2_iget(sb, no_addr);
        ip = GFS2_I(inode);
 
        if (!inode)
@@ -237,13 +230,100 @@ fail_glock:
 fail_iopen:
        gfs2_glock_put(io_gl);
 fail_put:
-       ip->i_gl->gl_object = NULL;
+       if (inode->i_state & I_NEW)
+               ip->i_gl->gl_object = NULL;
        gfs2_glock_put(ip->i_gl);
 fail:
-       iget_failed(inode);
+       if (inode->i_state & I_NEW)
+               iget_failed(inode);
+       else
+               iput(inode);
        return ERR_PTR(error);
 }
 
+/**
+ * gfs2_unlinked_inode_lookup - Lookup an unlinked inode for reclamation
+ * @sb: The super block
+ * no_addr: The inode number
+ * @@inode: A pointer to the inode found, if any
+ *
+ * Returns: 0 and *inode if no errors occurred.  If an error occurs,
+ *          the resulting *inode may or may not be NULL.
+ */
+
+int gfs2_unlinked_inode_lookup(struct super_block *sb, u64 no_addr,
+                              struct inode **inode)
+{
+       struct gfs2_sbd *sdp;
+       struct gfs2_inode *ip;
+       struct gfs2_glock *io_gl;
+       int error;
+       struct gfs2_holder gh;
+
+       *inode = gfs2_iget_skip(sb, no_addr);
+
+       if (!(*inode))
+               return -ENOBUFS;
+
+       if (!((*inode)->i_state & I_NEW))
+               return -ENOBUFS;
+
+       ip = GFS2_I(*inode);
+       sdp = GFS2_SB(*inode);
+       ip->i_no_formal_ino = -1;
+
+       error = gfs2_glock_get(sdp, no_addr, &gfs2_inode_glops, CREATE, &ip->i_gl);
+       if (unlikely(error))
+               goto fail;
+       ip->i_gl->gl_object = ip;
+
+       error = gfs2_glock_get(sdp, no_addr, &gfs2_iopen_glops, CREATE, &io_gl);
+       if (unlikely(error))
+               goto fail_put;
+
+       set_bit(GIF_INVALID, &ip->i_flags);
+       error = gfs2_glock_nq_init(io_gl, LM_ST_SHARED, LM_FLAG_TRY | GL_EXACT,
+                                  &ip->i_iopen_gh);
+       if (unlikely(error)) {
+               if (error == GLR_TRYFAILED)
+                       error = 0;
+               goto fail_iopen;
+       }
+       ip->i_iopen_gh.gh_gl->gl_object = ip;
+       gfs2_glock_put(io_gl);
+
+       (*inode)->i_mode = DT2IF(DT_UNKNOWN);
+
+       /*
+        * We must read the inode in order to work out its type in
+        * this case. Note that this doesn't happen often as we normally
+        * know the type beforehand. This code path only occurs during
+        * unlinked inode recovery (where it is safe to do this glock,
+        * which is not true in the general case).
+        */
+       error = gfs2_glock_nq_init(ip->i_gl, LM_ST_EXCLUSIVE, LM_FLAG_TRY,
+                                  &gh);
+       if (unlikely(error)) {
+               if (error == GLR_TRYFAILED)
+                       error = 0;
+               goto fail_glock;
+       }
+       /* Inode is now uptodate */
+       gfs2_glock_dq_uninit(&gh);
+       gfs2_set_iop(*inode);
+
+       return 0;
+fail_glock:
+       gfs2_glock_dq(&ip->i_iopen_gh);
+fail_iopen:
+       gfs2_glock_put(io_gl);
+fail_put:
+       ip->i_gl->gl_object = NULL;
+       gfs2_glock_put(ip->i_gl);
+fail:
+       return error;
+}
+
 static int gfs2_dinode_in(struct gfs2_inode *ip, const void *buf)
 {
        const struct gfs2_dinode *str = buf;
@@ -520,139 +600,6 @@ out:
        return inode ? inode : ERR_PTR(error);
 }
 
-static void gfs2_inum_range_in(struct gfs2_inum_range_host *ir, const void *buf)
-{
-       const struct gfs2_inum_range *str = buf;
-
-       ir->ir_start = be64_to_cpu(str->ir_start);
-       ir->ir_length = be64_to_cpu(str->ir_length);
-}
-
-static void gfs2_inum_range_out(const struct gfs2_inum_range_host *ir, void *buf)
-{
-       struct gfs2_inum_range *str = buf;
-
-       str->ir_start = cpu_to_be64(ir->ir_start);
-       str->ir_length = cpu_to_be64(ir->ir_length);
-}
-
-static int pick_formal_ino_1(struct gfs2_sbd *sdp, u64 *formal_ino)
-{
-       struct gfs2_inode *ip = GFS2_I(sdp->sd_ir_inode);
-       struct buffer_head *bh;
-       struct gfs2_inum_range_host ir;
-       int error;
-
-       error = gfs2_trans_begin(sdp, RES_DINODE, 0);
-       if (error)
-               return error;
-       mutex_lock(&sdp->sd_inum_mutex);
-
-       error = gfs2_meta_inode_buffer(ip, &bh);
-       if (error) {
-               mutex_unlock(&sdp->sd_inum_mutex);
-               gfs2_trans_end(sdp);
-               return error;
-       }
-
-       gfs2_inum_range_in(&ir, bh->b_data + sizeof(struct gfs2_dinode));
-
-       if (ir.ir_length) {
-               *formal_ino = ir.ir_start++;
-               ir.ir_length--;
-               gfs2_trans_add_bh(ip->i_gl, bh, 1);
-               gfs2_inum_range_out(&ir,
-                                   bh->b_data + sizeof(struct gfs2_dinode));
-               brelse(bh);
-               mutex_unlock(&sdp->sd_inum_mutex);
-               gfs2_trans_end(sdp);
-               return 0;
-       }
-
-       brelse(bh);
-
-       mutex_unlock(&sdp->sd_inum_mutex);
-       gfs2_trans_end(sdp);
-
-       return 1;
-}
-
-static int pick_formal_ino_2(struct gfs2_sbd *sdp, u64 *formal_ino)
-{
-       struct gfs2_inode *ip = GFS2_I(sdp->sd_ir_inode);
-       struct gfs2_inode *m_ip = GFS2_I(sdp->sd_inum_inode);
-       struct gfs2_holder gh;
-       struct buffer_head *bh;
-       struct gfs2_inum_range_host ir;
-       int error;
-
-       error = gfs2_glock_nq_init(m_ip->i_gl, LM_ST_EXCLUSIVE, 0, &gh);
-       if (error)
-               return error;
-
-       error = gfs2_trans_begin(sdp, 2 * RES_DINODE, 0);
-       if (error)
-               goto out;
-       mutex_lock(&sdp->sd_inum_mutex);
-
-       error = gfs2_meta_inode_buffer(ip, &bh);
-       if (error)
-               goto out_end_trans;
-
-       gfs2_inum_range_in(&ir, bh->b_data + sizeof(struct gfs2_dinode));
-
-       if (!ir.ir_length) {
-               struct buffer_head *m_bh;
-               u64 x, y;
-               __be64 z;
-
-               error = gfs2_meta_inode_buffer(m_ip, &m_bh);
-               if (error)
-                       goto out_brelse;
-
-               z = *(__be64 *)(m_bh->b_data + sizeof(struct gfs2_dinode));
-               x = y = be64_to_cpu(z);
-               ir.ir_start = x;
-               ir.ir_length = GFS2_INUM_QUANTUM;
-               x += GFS2_INUM_QUANTUM;
-               if (x < y)
-                       gfs2_consist_inode(m_ip);
-               z = cpu_to_be64(x);
-               gfs2_trans_add_bh(m_ip->i_gl, m_bh, 1);
-               *(__be64 *)(m_bh->b_data + sizeof(struct gfs2_dinode)) = z;
-
-               brelse(m_bh);
-       }
-
-       *formal_ino = ir.ir_start++;
-       ir.ir_length--;
-
-       gfs2_trans_add_bh(ip->i_gl, bh, 1);
-       gfs2_inum_range_out(&ir, bh->b_data + sizeof(struct gfs2_dinode));
-
-out_brelse:
-       brelse(bh);
-out_end_trans:
-       mutex_unlock(&sdp->sd_inum_mutex);
-       gfs2_trans_end(sdp);
-out:
-       gfs2_glock_dq_uninit(&gh);
-       return error;
-}
-
-static int pick_formal_ino(struct gfs2_sbd *sdp, u64 *inum)
-{
-       int error;
-
-       error = pick_formal_ino_1(sdp, inum);
-       if (error <= 0)
-               return error;
-
-       error = pick_formal_ino_2(sdp, inum);
-
-       return error;
-}
-
 /**
  * create_ok - OK to create a new on-disk inode here?
  * @dip:  Directory in which dinode is to be created
@@ -732,7 +679,7 @@ static int alloc_dinode(struct gfs2_inode *dip, u64 *no_addr, u64 *generation)
        if (error)
                goto out_ipreserv;
 
-       *no_addr = gfs2_alloc_di(dip, generation);
+       error = gfs2_alloc_di(dip, no_addr, generation);
 
        gfs2_trans_end(sdp);
 
@@ -925,7 +872,6 @@ static int gfs2_security_init(struct gfs2_inode *dip, struct gfs2_inode *ip)
        size_t len;
        void *value;
        char *name;
-       struct gfs2_ea_request er;
 
        err = security_inode_init_security(&ip->i_inode, &dip->i_inode,
                                           &name, &value, &len);
@@ -936,16 +882,8 @@ static int gfs2_security_init(struct gfs2_inode *dip, struct gfs2_inode *ip)
                return err;
        }
 
-       memset(&er, 0, sizeof(struct gfs2_ea_request));
-
-       er.er_type = GFS2_EATYPE_SECURITY;
-       er.er_name = name;
-       er.er_data = value;
-       er.er_name_len = strlen(name);
-       er.er_data_len = len;
-
-       err = gfs2_ea_set_i(ip, &er);
-
+       err = __gfs2_xattr_set(&ip->i_inode, name, value, len, 0,
+                              GFS2_EATYPE_SECURITY);
        kfree(value);
        kfree(name);
 
@@ -992,13 +930,10 @@ struct inode *gfs2_createi(struct gfs2_holder *ghs, const struct qstr *name,
        if (error)
                goto fail_gunlock;
 
-       error = pick_formal_ino(sdp, &inum.no_formal_ino);
-       if (error)
-               goto fail_gunlock;
-
        error = alloc_dinode(dip, &inum.no_addr, &generation);
        if (error)
                goto fail_gunlock;
+       inum.no_formal_ino = generation;
 
        error = gfs2_glock_nq_num(sdp, inum.no_addr, &gfs2_inode_glops,
                                  LM_ST_EXCLUSIVE, GL_SKIP, ghs + 1);
@@ -1009,9 +944,8 @@ struct inode *gfs2_createi(struct gfs2_holder *ghs, const struct qstr *name,
        if (error)
                goto fail_gunlock2;
 
-       inode = gfs2_inode_lookup(dir->i_sb, IF2DT(mode),
-                                       inum.no_addr,
-                                       inum.no_formal_ino, 0);
+       inode = gfs2_inode_lookup(dir->i_sb, IF2DT(mode), inum.no_addr,
+                                 inum.no_formal_ino);
        if (IS_ERR(inode))
                goto fail_gunlock2;
 
@@ -1019,7 +953,7 @@ struct inode *gfs2_createi(struct gfs2_holder *ghs, const struct qstr *name,
        if (error)
                goto fail_gunlock2;
 
-       error = gfs2_acl_create(dip, GFS2_I(inode));
+       error = gfs2_acl_create(dip, inode);
        if (error)
                goto fail_gunlock2;
 
@@ -1047,154 +981,7 @@ fail:
        return ERR_PTR(error);
 }
 
-/**
- * gfs2_rmdiri - Remove a directory
- * @dip: The parent directory of the directory to be removed
- * @name: The name of the directory to be removed
- * @ip: The GFS2 inode of the directory to be removed
- *
- * Assumes Glocks on dip and ip are held
- *
- * Returns: errno
- */
-
-int gfs2_rmdiri(struct gfs2_inode *dip, const struct qstr *name,
-               struct gfs2_inode *ip)
-{
-       struct qstr dotname;
-       int error;
-
-       if (ip->i_entries != 2) {
-               if (gfs2_consist_inode(ip))
-                       gfs2_dinode_print(ip);
-               return -EIO;
-       }
-
-       error = gfs2_dir_del(dip, name);
-       if (error)
-               return error;
-
-       error = gfs2_change_nlink(dip, -1);
-       if (error)
-               return error;
-
-       gfs2_str2qstr(&dotname, ".");
-       error = gfs2_dir_del(ip, &dotname);
-       if (error)
-               return error;
-
-       gfs2_str2qstr(&dotname, "..");
-       error = gfs2_dir_del(ip, &dotname);
-       if (error)
-               return error;
-
-       /* It looks odd, but it really should be done twice */
-       error = gfs2_change_nlink(ip, -1);
-       if (error)
-               return error;
-
-       error = gfs2_change_nlink(ip, -1);
-       if (error)
-               return error;
-
-       return error;
-}
-
-/*
- * gfs2_unlink_ok - check to see that a inode is still in a directory
- * @dip: the directory
- * @name: the name of the file
- * @ip: the inode
- *
- * Assumes that the lock on (at least) @dip is held.
- *
- * Returns: 0 if the parent/child relationship is correct, errno if it isn't
- */
-
-int gfs2_unlink_ok(struct gfs2_inode *dip, const struct qstr *name,
-                  const struct gfs2_inode *ip)
-{
-       int error;
-
-       if (IS_IMMUTABLE(&ip->i_inode) || IS_APPEND(&ip->i_inode))
-               return -EPERM;
-
-       if ((dip->i_inode.i_mode & S_ISVTX) &&
-           dip->i_inode.i_uid != current_fsuid() &&
-           ip->i_inode.i_uid != current_fsuid() && !capable(CAP_FOWNER))
-               return -EPERM;
-
-       if (IS_APPEND(&dip->i_inode))
-               return -EPERM;
-
-       error = gfs2_permission(&dip->i_inode, MAY_WRITE | MAY_EXEC);
-       if (error)
-               return error;
-
-       error = gfs2_dir_check(&dip->i_inode, name, ip);
-       if (error)
-               return error;
-
-       return 0;
-}
-
-/**
- * gfs2_readlinki - return the contents of a symlink
- * @ip: the symlink's inode
- * @buf: a pointer to the buffer to be filled
- * @len: a pointer to the length of @buf
- *
- * If @buf is too small, a piece of memory is kmalloc()ed and needs
- * to be freed by the caller.
- *
- * Returns: errno
- */
-
-int gfs2_readlinki(struct gfs2_inode *ip, char **buf, unsigned int *len)
-{
-       struct gfs2_holder i_gh;
-       struct buffer_head *dibh;
-       unsigned int x;
-       int error;
-
-       gfs2_holder_init(ip->i_gl, LM_ST_SHARED, 0, &i_gh);
-       error = gfs2_glock_nq(&i_gh);
-       if (error) {
-               gfs2_holder_uninit(&i_gh);
-               return error;
-       }
-
-       if (!ip->i_disksize) {
-               gfs2_consist_inode(ip);
-               error = -EIO;
-               goto out;
-       }
-
-       error = gfs2_meta_inode_buffer(ip, &dibh);
-       if (error)
-               goto out;
-
-       x = ip->i_disksize + 1;
-       if (x > *len) {
-               *buf = kmalloc(x, GFP_NOFS);
-               if (!*buf) {
-                       error = -ENOMEM;
-                       goto out_brelse;
-               }
-       }
-
-       memcpy(*buf, dibh->b_data + sizeof(struct gfs2_dinode), x);
-       *len = x;
-
-out_brelse:
-       brelse(dibh);
-out:
-       gfs2_glock_dq_uninit(&i_gh);
-       return error;
-}
-
-static int
-__gfs2_setattr_simple(struct gfs2_inode *ip, struct iattr *attr)
+static int __gfs2_setattr_simple(struct gfs2_inode *ip, struct iattr *attr)
 {
        struct buffer_head *dibh;
        int error;
@@ -1242,9 +1029,7 @@ void gfs2_dinode_out(const struct gfs2_inode *ip, void *buf)
 
        str->di_header.mh_magic = cpu_to_be32(GFS2_MAGIC);
        str->di_header.mh_type = cpu_to_be32(GFS2_METATYPE_DI);
-       str->di_header.__pad0 = 0;
        str->di_header.mh_format = cpu_to_be32(GFS2_FORMAT_DI);
-       str->di_header.__pad1 = 0;
        str->di_num.no_addr = cpu_to_be64(ip->i_no_addr);
        str->di_num.no_formal_ino = cpu_to_be64(ip->i_no_formal_ino);
        str->di_mode = cpu_to_be32(ip->i_inode.i_mode);