]> git.baikalelectronics.ru Git - kernel.git/commitdiff
9p fid refcount: add p9_fid_get/put wrappers
authorDominique Martinet <asmadeus@codewreck.org>
Sun, 12 Jun 2022 04:42:32 +0000 (13:42 +0900)
committerDominique Martinet <dominique.martinet@atmark-techno.com>
Sat, 2 Jul 2022 09:52:21 +0000 (18:52 +0900)
I was recently reminded that it is not clear that p9_client_clunk()
was actually just decrementing refcount and clunking only when that
reaches zero: make it clear through a set of helpers.

This will also allow instrumenting refcounting better for debugging
next patch

Link: https://lkml.kernel.org/r/20220612085330.1451496-5-asmadeus@codewreck.org
Reviewed-by: Tyler Hicks <tyhicks@linux.microsoft.com>
Reviewed-by: Christian Schoenebeck <linux_oss@crudebyte.com>
Signed-off-by: Dominique Martinet <asmadeus@codewreck.org>
12 files changed:
fs/9p/fid.c
fs/9p/fid.h
fs/9p/vfs_addr.c
fs/9p/vfs_dentry.c
fs/9p/vfs_dir.c
fs/9p/vfs_file.c
fs/9p/vfs_inode.c
fs/9p/vfs_inode_dotl.c
fs/9p/vfs_super.c
fs/9p/xattr.c
include/net/9p/client.h
net/9p/client.c

index e8fad28fc5bdb61652fee563df16d581bc21d5fe..d792499349c49be0063bb8b4e34c37d311f6350a 100644 (file)
@@ -56,7 +56,7 @@ static struct p9_fid *v9fs_fid_find_inode(struct inode *inode, kuid_t uid)
        h = (struct hlist_head *)&inode->i_private;
        hlist_for_each_entry(fid, h, ilist) {
                if (uid_eq(fid->uid, uid)) {
-                       refcount_inc(&fid->count);
+                       p9_fid_get(fid);
                        ret = fid;
                        break;
                }
@@ -104,7 +104,7 @@ static struct p9_fid *v9fs_fid_find(struct dentry *dentry, kuid_t uid, int any)
                hlist_for_each_entry(fid, h, dlist) {
                        if (any || uid_eq(fid->uid, uid)) {
                                ret = fid;
-                               refcount_inc(&ret->count);
+                               p9_fid_get(ret);
                                break;
                        }
                }
@@ -172,7 +172,7 @@ static struct p9_fid *v9fs_fid_lookup_with_uid(struct dentry *dentry,
                old_fid = fid;
 
                fid = p9_client_walk(old_fid, 1, &dentry->d_name.name, 1);
-               p9_client_clunk(old_fid);
+               p9_fid_put(old_fid);
                goto fid_out;
        }
        up_read(&v9ses->rename_sem);
@@ -194,7 +194,7 @@ static struct p9_fid *v9fs_fid_lookup_with_uid(struct dentry *dentry,
                if (IS_ERR(root_fid))
                        return root_fid;
 
-               refcount_inc(&root_fid->count);
+               p9_fid_get(root_fid);
                v9fs_fid_add(dentry->d_sb->s_root, root_fid);
        }
        /* If we are root ourself just return that */
@@ -225,7 +225,7 @@ static struct p9_fid *v9fs_fid_lookup_with_uid(struct dentry *dentry,
                                     old_fid == root_fid /* clone */);
                /* non-cloning walk will return the same fid */
                if (fid != old_fid) {
-                       p9_client_clunk(old_fid);
+                       p9_fid_put(old_fid);
                        old_fid = fid;
                }
                if (IS_ERR(fid)) {
@@ -240,11 +240,11 @@ fid_out:
                spin_lock(&dentry->d_lock);
                if (d_unhashed(dentry)) {
                        spin_unlock(&dentry->d_lock);
-                       p9_client_clunk(fid);
+                       p9_fid_put(fid);
                        fid = ERR_PTR(-ENOENT);
                } else {
                        __add_fid(dentry, fid);
-                       refcount_inc(&fid->count);
+                       p9_fid_get(fid);
                        spin_unlock(&dentry->d_lock);
                }
        }
@@ -301,7 +301,7 @@ struct p9_fid *v9fs_writeback_fid(struct dentry *dentry)
        fid = clone_fid(ofid);
        if (IS_ERR(fid))
                goto error_out;
-       p9_client_clunk(ofid);
+       p9_fid_put(ofid);
        /*
         * writeback fid will only be used to write back the
         * dirty pages. We always request for the open fid in read-write
@@ -310,7 +310,7 @@ struct p9_fid *v9fs_writeback_fid(struct dentry *dentry)
         */
        err = p9_client_open(fid, O_RDWR);
        if (err < 0) {
-               p9_client_clunk(fid);
+               p9_fid_put(fid);
                fid = ERR_PTR(err);
                goto error_out;
        }
index f7f33509e1692acbbb87d0f28bc7e757ec3db394..3168dfad510eca01a64b94f2c11f03f0c8ad91e9 100644 (file)
@@ -29,7 +29,7 @@ static inline struct p9_fid *v9fs_fid_clone(struct dentry *dentry)
                return fid;
 
        nfid = clone_fid(fid);
-       p9_client_clunk(fid);
+       p9_fid_put(fid);
        return nfid;
 }
 #endif
index d0833fa69faf1ba01d2dfdff47728a588b0d6a6d..47b9a1122f344f304623c4d66ab7e4f84e356de6 100644 (file)
@@ -73,7 +73,7 @@ static int v9fs_init_request(struct netfs_io_request *rreq, struct file *file)
                BUG_ON(!fid);
        }
 
-       refcount_inc(&fid->count);
+       p9_fid_get(fid);
        rreq->netfs_priv = fid;
        return 0;
 }
@@ -86,7 +86,7 @@ static void v9fs_free_request(struct netfs_io_request *rreq)
 {
        struct p9_fid *fid = rreq->netfs_priv;
 
-       p9_client_clunk(fid);
+       p9_fid_put(fid);
 }
 
 /**
index 1c609e99d280f8c17ac0734d559ce383e3b3e3aa..f89f01734587b2e9a8f31c8002dbdce7b2d910d3 100644 (file)
@@ -54,7 +54,7 @@ static void v9fs_dentry_release(struct dentry *dentry)
        p9_debug(P9_DEBUG_VFS, " dentry: %pd (%p)\n",
                 dentry, dentry);
        hlist_for_each_safe(p, n, (struct hlist_head *)&dentry->d_fsdata)
-               p9_client_clunk(hlist_entry(p, struct p9_fid, dlist));
+               p9_fid_put(hlist_entry(p, struct p9_fid, dlist));
        dentry->d_fsdata = NULL;
 }
 
@@ -85,7 +85,7 @@ static int v9fs_lookup_revalidate(struct dentry *dentry, unsigned int flags)
                        retval = v9fs_refresh_inode_dotl(fid, inode);
                else
                        retval = v9fs_refresh_inode(fid, inode);
-               p9_client_clunk(fid);
+               p9_fid_put(fid);
 
                if (retval == -ENOENT)
                        return 0;
index 958680f7f23e79a9f189ecc05675fd4688e84c41..000fbaae9b180ce749ae3c10eab6b56cfb98a0a6 100644 (file)
@@ -218,7 +218,7 @@ int v9fs_dir_release(struct inode *inode, struct file *filp)
                spin_lock(&inode->i_lock);
                hlist_del(&fid->ilist);
                spin_unlock(&inode->i_lock);
-               p9_client_clunk(fid);
+               p9_fid_put(fid);
        }
 
        if ((filp->f_mode & FMODE_WRITE)) {
index 2573c08f335c5d2d8057cbcb526d6abf68799904..8276f3af35d7911bd650cd5f1fb8f9ffe60621b9 100644 (file)
@@ -63,7 +63,7 @@ int v9fs_file_open(struct inode *inode, struct file *file)
 
                err = p9_client_open(fid, omode);
                if (err < 0) {
-                       p9_client_clunk(fid);
+                       p9_fid_put(fid);
                        return err;
                }
                if ((file->f_flags & O_APPEND) &&
@@ -98,7 +98,7 @@ int v9fs_file_open(struct inode *inode, struct file *file)
        v9fs_open_fid_add(inode, fid);
        return 0;
 out_error:
-       p9_client_clunk(file->private_data);
+       p9_fid_put(file->private_data);
        file->private_data = NULL;
        return err;
 }
index 3d8297714772cc6b91c7e74c1611f07fd4e59251..bde18776f0c77f8b8abd35bf1f8e63216da910f2 100644 (file)
@@ -400,7 +400,7 @@ void v9fs_evict_inode(struct inode *inode)
        fscache_relinquish_cookie(v9fs_inode_cookie(v9inode), false);
        /* clunk the fid stashed in writeback_fid */
        if (v9inode->writeback_fid) {
-               p9_client_clunk(v9inode->writeback_fid);
+               p9_fid_put(v9inode->writeback_fid);
                v9inode->writeback_fid = NULL;
        }
 }
@@ -569,7 +569,7 @@ static int v9fs_remove(struct inode *dir, struct dentry *dentry, int flags)
        if (v9fs_proto_dotl(v9ses))
                retval = p9_client_unlinkat(dfid, dentry->d_name.name,
                                            v9fs_at_to_dotl_flags(flags));
-       p9_client_clunk(dfid);
+       p9_fid_put(dfid);
        if (retval == -EOPNOTSUPP) {
                /* Try the one based on path */
                v9fid = v9fs_fid_clone(dentry);
@@ -633,14 +633,14 @@ v9fs_create(struct v9fs_session_info *v9ses, struct inode *dir,
        if (IS_ERR(ofid)) {
                err = PTR_ERR(ofid);
                p9_debug(P9_DEBUG_VFS, "p9_client_walk failed %d\n", err);
-               p9_client_clunk(dfid);
+               p9_fid_put(dfid);
                return ERR_PTR(err);
        }
 
        err = p9_client_fcreate(ofid, name, perm, mode, extension);
        if (err < 0) {
                p9_debug(P9_DEBUG_VFS, "p9_client_fcreate failed %d\n", err);
-               p9_client_clunk(dfid);
+               p9_fid_put(dfid);
                goto error;
        }
 
@@ -652,7 +652,7 @@ v9fs_create(struct v9fs_session_info *v9ses, struct inode *dir,
                        p9_debug(P9_DEBUG_VFS,
                                   "p9_client_walk failed %d\n", err);
                        fid = NULL;
-                       p9_client_clunk(dfid);
+                       p9_fid_put(dfid);
                        goto error;
                }
                /*
@@ -663,20 +663,20 @@ v9fs_create(struct v9fs_session_info *v9ses, struct inode *dir,
                        err = PTR_ERR(inode);
                        p9_debug(P9_DEBUG_VFS,
                                   "inode creation failed %d\n", err);
-                       p9_client_clunk(dfid);
+                       p9_fid_put(dfid);
                        goto error;
                }
                v9fs_fid_add(dentry, fid);
                d_instantiate(dentry, inode);
        }
-       p9_client_clunk(dfid);
+       p9_fid_put(dfid);
        return ofid;
 error:
        if (ofid)
-               p9_client_clunk(ofid);
+               p9_fid_put(ofid);
 
        if (fid)
-               p9_client_clunk(fid);
+               p9_fid_put(fid);
 
        return ERR_PTR(err);
 }
@@ -708,7 +708,7 @@ v9fs_vfs_create(struct user_namespace *mnt_userns, struct inode *dir,
                return PTR_ERR(fid);
 
        v9fs_invalidate_inode_attr(dir);
-       p9_client_clunk(fid);
+       p9_fid_put(fid);
 
        return 0;
 }
@@ -744,7 +744,7 @@ static int v9fs_vfs_mkdir(struct user_namespace *mnt_userns, struct inode *dir,
        }
 
        if (fid)
-               p9_client_clunk(fid);
+               p9_fid_put(fid);
 
        return err;
 }
@@ -785,7 +785,7 @@ struct dentry *v9fs_vfs_lookup(struct inode *dir, struct dentry *dentry,
         */
        name = dentry->d_name.name;
        fid = p9_client_walk(dfid, 1, &name, 1);
-       p9_client_clunk(dfid);
+       p9_fid_put(dfid);
        if (fid == ERR_PTR(-ENOENT))
                inode = NULL;
        else if (IS_ERR(fid))
@@ -808,7 +808,7 @@ struct dentry *v9fs_vfs_lookup(struct inode *dir, struct dentry *dentry,
                else if (!IS_ERR(res))
                        v9fs_fid_add(res, fid);
                else
-                       p9_client_clunk(fid);
+                       p9_fid_put(fid);
        }
        return res;
 }
@@ -891,7 +891,7 @@ out:
 
 error:
        if (fid)
-               p9_client_clunk(fid);
+               p9_fid_put(fid);
        goto out;
 }
 
@@ -959,7 +959,7 @@ v9fs_vfs_rename(struct user_namespace *mnt_userns, struct inode *old_dir,
        dfid = v9fs_parent_fid(old_dentry);
        olddirfid = clone_fid(dfid);
        if (dfid && !IS_ERR(dfid))
-               p9_client_clunk(dfid);
+               p9_fid_put(dfid);
 
        if (IS_ERR(olddirfid)) {
                retval = PTR_ERR(olddirfid);
@@ -968,7 +968,7 @@ v9fs_vfs_rename(struct user_namespace *mnt_userns, struct inode *old_dir,
 
        dfid = v9fs_parent_fid(new_dentry);
        newdirfid = clone_fid(dfid);
-       p9_client_clunk(dfid);
+       p9_fid_put(dfid);
 
        if (IS_ERR(newdirfid)) {
                retval = PTR_ERR(newdirfid);
@@ -1020,13 +1020,13 @@ clunk_newdir:
                d_move(old_dentry, new_dentry);
        }
        up_write(&v9ses->rename_sem);
-       p9_client_clunk(newdirfid);
+       p9_fid_put(newdirfid);
 
 clunk_olddir:
-       p9_client_clunk(olddirfid);
+       p9_fid_put(olddirfid);
 
 done:
-       p9_client_clunk(oldfid);
+       p9_fid_put(oldfid);
        return retval;
 }
 
@@ -1060,7 +1060,7 @@ v9fs_vfs_getattr(struct user_namespace *mnt_userns, const struct path *path,
                return PTR_ERR(fid);
 
        st = p9_client_stat(fid);
-       p9_client_clunk(fid);
+       p9_fid_put(fid);
        if (IS_ERR(st))
                return PTR_ERR(st);
 
@@ -1136,7 +1136,7 @@ static int v9fs_vfs_setattr(struct user_namespace *mnt_userns,
        retval = p9_client_wstat(fid, &wstat);
 
        if (use_dentry)
-               p9_client_clunk(fid);
+               p9_fid_put(fid);
 
        if (retval < 0)
                return retval;
@@ -1261,7 +1261,7 @@ static const char *v9fs_vfs_get_link(struct dentry *dentry,
                return ERR_CAST(fid);
 
        st = p9_client_stat(fid);
-       p9_client_clunk(fid);
+       p9_fid_put(fid);
        if (IS_ERR(st))
                return ERR_CAST(st);
 
@@ -1308,7 +1308,7 @@ static int v9fs_vfs_mkspecial(struct inode *dir, struct dentry *dentry,
                return PTR_ERR(fid);
 
        v9fs_invalidate_inode_attr(dir);
-       p9_client_clunk(fid);
+       p9_fid_put(fid);
        return 0;
 }
 
@@ -1364,7 +1364,7 @@ v9fs_vfs_link(struct dentry *old_dentry, struct inode *dir,
                v9fs_refresh_inode(oldfid, d_inode(old_dentry));
                v9fs_invalidate_inode_attr(dir);
        }
-       p9_client_clunk(oldfid);
+       p9_fid_put(oldfid);
        return retval;
 }
 
index b6eb1160296c3033d5e0a2347b1bbf8890c797c7..09b124fe349caf4c48438ef5d1c61ac096b02871 100644 (file)
@@ -274,7 +274,7 @@ v9fs_vfs_atomic_open_dotl(struct inode *dir, struct dentry *dentry,
        if (IS_ERR(ofid)) {
                err = PTR_ERR(ofid);
                p9_debug(P9_DEBUG_VFS, "p9_client_walk failed %d\n", err);
-               p9_client_clunk(dfid);
+               p9_fid_put(dfid);
                goto out;
        }
 
@@ -286,7 +286,7 @@ v9fs_vfs_atomic_open_dotl(struct inode *dir, struct dentry *dentry,
        if (err) {
                p9_debug(P9_DEBUG_VFS, "Failed to get acl values in creat %d\n",
                         err);
-               p9_client_clunk(dfid);
+               p9_fid_put(dfid);
                goto error;
        }
        err = p9_client_create_dotl(ofid, name, v9fs_open_to_dotl_flags(flags),
@@ -294,14 +294,14 @@ v9fs_vfs_atomic_open_dotl(struct inode *dir, struct dentry *dentry,
        if (err < 0) {
                p9_debug(P9_DEBUG_VFS, "p9_client_open_dotl failed in creat %d\n",
                         err);
-               p9_client_clunk(dfid);
+               p9_fid_put(dfid);
                goto error;
        }
        v9fs_invalidate_inode_attr(dir);
 
        /* instantiate inode and assign the unopened fid to the dentry */
        fid = p9_client_walk(dfid, 1, &name, 1);
-       p9_client_clunk(dfid);
+       p9_fid_put(dfid);
        if (IS_ERR(fid)) {
                err = PTR_ERR(fid);
                p9_debug(P9_DEBUG_VFS, "p9_client_walk failed %d\n", err);
@@ -358,10 +358,10 @@ out:
 
 error:
        if (fid)
-               p9_client_clunk(fid);
+               p9_fid_put(fid);
 err_clunk_old_fid:
        if (ofid)
-               p9_client_clunk(ofid);
+               p9_fid_put(ofid);
        goto out;
 }
 
@@ -458,9 +458,9 @@ static int v9fs_vfs_mkdir_dotl(struct user_namespace *mnt_userns,
        v9fs_invalidate_inode_attr(dir);
 error:
        if (fid)
-               p9_client_clunk(fid);
+               p9_fid_put(fid);
        v9fs_put_acl(dacl, pacl);
-       p9_client_clunk(dfid);
+       p9_fid_put(dfid);
        return err;
 }
 
@@ -489,7 +489,7 @@ v9fs_vfs_getattr_dotl(struct user_namespace *mnt_userns,
         */
 
        st = p9_client_getattr_dotl(fid, P9_STATS_ALL);
-       p9_client_clunk(fid);
+       p9_fid_put(fid);
        if (IS_ERR(st))
                return PTR_ERR(st);
 
@@ -603,7 +603,7 @@ int v9fs_vfs_setattr_dotl(struct user_namespace *mnt_userns,
        retval = p9_client_setattr(fid, &p9attr);
        if (retval < 0) {
                if (use_dentry)
-                       p9_client_clunk(fid);
+                       p9_fid_put(fid);
                return retval;
        }
 
@@ -619,12 +619,12 @@ int v9fs_vfs_setattr_dotl(struct user_namespace *mnt_userns,
                retval = v9fs_acl_chmod(inode, fid);
                if (retval < 0) {
                        if (use_dentry)
-                               p9_client_clunk(fid);
+                               p9_fid_put(fid);
                        return retval;
                }
        }
        if (use_dentry)
-               p9_client_clunk(fid);
+               p9_fid_put(fid);
 
        return 0;
 }
@@ -771,9 +771,9 @@ v9fs_vfs_symlink_dotl(struct user_namespace *mnt_userns, struct inode *dir,
 
 error:
        if (fid)
-               p9_client_clunk(fid);
+               p9_fid_put(fid);
 
-       p9_client_clunk(dfid);
+       p9_fid_put(dfid);
        return err;
 }
 
@@ -803,14 +803,14 @@ v9fs_vfs_link_dotl(struct dentry *old_dentry, struct inode *dir,
 
        oldfid = v9fs_fid_lookup(old_dentry);
        if (IS_ERR(oldfid)) {
-               p9_client_clunk(dfid);
+               p9_fid_put(dfid);
                return PTR_ERR(oldfid);
        }
 
        err = p9_client_link(dfid, oldfid, dentry->d_name.name);
 
-       p9_client_clunk(dfid);
-       p9_client_clunk(oldfid);
+       p9_fid_put(dfid);
+       p9_fid_put(oldfid);
        if (err < 0) {
                p9_debug(P9_DEBUG_VFS, "p9_client_link failed %d\n", err);
                return err;
@@ -826,7 +826,7 @@ v9fs_vfs_link_dotl(struct dentry *old_dentry, struct inode *dir,
                        return PTR_ERR(fid);
 
                v9fs_refresh_inode_dotl(fid, d_inode(old_dentry));
-               p9_client_clunk(fid);
+               p9_fid_put(fid);
        }
        ihold(d_inode(old_dentry));
        d_instantiate(dentry, d_inode(old_dentry));
@@ -924,9 +924,9 @@ v9fs_vfs_mknod_dotl(struct user_namespace *mnt_userns, struct inode *dir,
        }
 error:
        if (fid)
-               p9_client_clunk(fid);
+               p9_fid_put(fid);
        v9fs_put_acl(dacl, pacl);
-       p9_client_clunk(dfid);
+       p9_fid_put(dfid);
 
        return err;
 }
@@ -956,7 +956,7 @@ v9fs_vfs_get_link_dotl(struct dentry *dentry,
        if (IS_ERR(fid))
                return ERR_CAST(fid);
        retval = p9_client_readlink(fid, &target);
-       p9_client_clunk(fid);
+       p9_fid_put(fid);
        if (retval)
                return ERR_PTR(retval);
        set_delayed_call(done, kfree_link, target);
index 97e23b4e69820200c5c4859588168761d97d12f8..bf350fad95007bf0b19201ab47dd23aa9a33bdee 100644 (file)
@@ -190,7 +190,7 @@ static struct dentry *v9fs_mount(struct file_system_type *fs_type, int flags,
        return dget(sb->s_root);
 
 clunk_fid:
-       p9_client_clunk(fid);
+       p9_fid_put(fid);
        v9fs_session_close(v9ses);
 free_session:
        kfree(v9ses);
@@ -203,7 +203,7 @@ release_sb:
         * attached the fid to dentry so it won't get clunked
         * automatically.
         */
-       p9_client_clunk(fid);
+       p9_fid_put(fid);
        deactivate_locked_super(sb);
        return ERR_PTR(retval);
 }
@@ -270,7 +270,7 @@ static int v9fs_statfs(struct dentry *dentry, struct kstatfs *buf)
        }
        res = simple_statfs(dentry, buf);
 done:
-       p9_client_clunk(fid);
+       p9_fid_put(fid);
        return res;
 }
 
index a824441b95a2721cb1738630381799c867553b5d..1f9298a4bd4288390c1e98395486931ff4c4ac93 100644 (file)
@@ -44,7 +44,7 @@ ssize_t v9fs_fid_xattr_get(struct p9_fid *fid, const char *name,
                if (err)
                        retval = err;
        }
-       p9_client_clunk(attr_fid);
+       p9_fid_put(attr_fid);
        return retval;
 }
 
@@ -71,7 +71,7 @@ ssize_t v9fs_xattr_get(struct dentry *dentry, const char *name,
        if (IS_ERR(fid))
                return PTR_ERR(fid);
        ret = v9fs_fid_xattr_get(fid, name, buffer, buffer_size);
-       p9_client_clunk(fid);
+       p9_fid_put(fid);
 
        return ret;
 }
@@ -98,7 +98,7 @@ int v9fs_xattr_set(struct dentry *dentry, const char *name,
        if (IS_ERR(fid))
                return PTR_ERR(fid);
        ret = v9fs_fid_xattr_set(fid, name, value, value_len, flags);
-       p9_client_clunk(fid);
+       p9_fid_put(fid);
        return ret;
 }
 
@@ -128,7 +128,7 @@ int v9fs_fid_xattr_set(struct p9_fid *fid, const char *name,
                         retval);
        else
                p9_client_write(fid, 0, &from, &retval);
-       err = p9_client_clunk(fid);
+       err = p9_fid_put(fid);
        if (!retval && err)
                retval = err;
        return retval;
index ec1d1706f43c03fad268f7a92f47b2abb7170374..eabb53992350c521d523b0ff87ac471ec059125f 100644 (file)
@@ -237,6 +237,34 @@ static inline int p9_req_try_get(struct p9_req_t *r)
 
 int p9_req_put(struct p9_req_t *r);
 
+/* fid reference counting helpers:
+ *  - fids used for any length of time should always be referenced through
+ *    p9_fid_get(), and released with p9_fid_put()
+ *  - v9fs_fid_lookup() or similar will automatically call get for you
+ *    and also require a put
+ *  - the *_fid_add() helpers will stash the fid in the inode,
+ *    at which point it is the responsibility of evict_inode()
+ *    to call the put
+ *  - the last put will automatically send a clunk to the server
+ */
+static inline struct p9_fid *p9_fid_get(struct p9_fid *fid)
+{
+       refcount_inc(&fid->count);
+
+       return fid;
+}
+
+static inline int p9_fid_put(struct p9_fid *fid)
+{
+       if (!fid || IS_ERR(fid))
+               return 0;
+
+       if (!refcount_dec_and_test(&fid->count))
+               return 0;
+
+       return p9_client_clunk(fid);
+}
+
 void p9_client_cb(struct p9_client *c, struct p9_req_t *req, int status);
 
 int p9_parse_header(struct p9_fcall *pdu, int32_t *size, int8_t *type,
index 8bba0d9cf975417b64b0483849f9201bd7fd4d2a..f3eb280c7d9d93a663f5a64b1cf40ec649e8372c 100644 (file)
@@ -1228,7 +1228,7 @@ struct p9_fid *p9_client_walk(struct p9_fid *oldfid, uint16_t nwname,
 
 clunk_fid:
        kfree(wqids);
-       p9_client_clunk(fid);
+       p9_fid_put(fid);
        fid = NULL;
 
 error:
@@ -1459,15 +1459,6 @@ int p9_client_clunk(struct p9_fid *fid)
        struct p9_req_t *req;
        int retries = 0;
 
-       if (!fid || IS_ERR(fid)) {
-               pr_warn("%s (%d): Trying to clunk with invalid fid\n",
-                       __func__, task_pid_nr(current));
-               dump_stack();
-               return 0;
-       }
-       if (!refcount_dec_and_test(&fid->count))
-               return 0;
-
 again:
        p9_debug(P9_DEBUG_9P, ">>> TCLUNK fid %d (try %d)\n",
                 fid->fid, retries);
@@ -1519,7 +1510,7 @@ int p9_client_remove(struct p9_fid *fid)
        p9_tag_remove(clnt, req);
 error:
        if (err == -ERESTARTSYS)
-               p9_client_clunk(fid);
+               p9_fid_put(fid);
        else
                p9_fid_destroy(fid);
        return err;
@@ -2042,7 +2033,7 @@ struct p9_fid *p9_client_xattrwalk(struct p9_fid *file_fid,
                 attr_fid->fid, *attr_size);
        return attr_fid;
 clunk_fid:
-       p9_client_clunk(attr_fid);
+       p9_fid_put(attr_fid);
        attr_fid = NULL;
 error:
        if (attr_fid && attr_fid != file_fid)