Whamcloud Community JIRA

Whamcloud Community JIRA https://jira.whamcloud.com This file is an XML representation of an issue en-us

9.4.14

940014

05-12-2023

[LU-6794] memory leak in Lustre NFS support code, LASSERT() at unmount https://jira.whamcloud.com/browse/LU-6794 Lustre <p>There is an unbalanced iput() call in Lustre NFS support code.</p> <p>A possible symptom is a kernel crash with<br/> ASSERTION( atomic_read(&d->ld_ref) == 0 ) failed</p> <p>A fix will be uploaded shortly.</p> LU-6794

memory leak in Lustre NFS support code, LASSERT() at unmount

Bug Critical Resolved Fixed Zhenyu Xu Andrew Perepechko patch Thu, 2 Jul 2015 20:22:14 +0000 Mon, 14 Sep 2015 14:37:28 +0000 Thu, 16 Jul 2015 13:00:25 +0000 Lustre 2.8.0 Lustre 2.8.0 0 4 <p>Andrew Perepechko (andrew.perepechko@seagate.com) uploaded a new patch: <a href="http://review.whamcloud.com/15480" class="external-link" target="_blank" rel="nofollow noopener">http://review.whamcloud.com/15480</a><br/> Subject: <a href="https://jira.whamcloud.com/browse/LU-6794" title="memory leak in Lustre NFS support code, LASSERT() at unmount" class="issue-link" data-issue-key="LU-6794"><del>LU-6794</del></a> nfs: ASSERTION( atomic_read(&d->ld_ref) == 0 ) failed<br/> Project: fs/lustre-release<br/> Branch: master<br/> Current Patch Set: 1<br/> Commit: 43a8fb75c98ba0d677eaa1297e21e0a4de2ec16f</p> <p>Bobijam</p> <p>Could you please take care of this patch?</p> <p>Thanks</p> <p>Peter</p> <p>Based on vanilla linux.git, d_obtain_alias() has had the feature of releasing inode on error from its introduction up to now:</p> <div class="code panel" style="border-width: 1px;"><div class="codeContent panelContent"> <pre class="code-java">commit 4ea3ada2955e4519befa98ff55dd62d6dfbd1705 Author: Christoph Hellwig <hch@lst.de> Date: Mon Aug 11 15:48:57 2008 +0200 [PATCH] <span class="code-keyword">new</span> helper: d_obtain_alias The calling conventions of d_alloc_anon are rather unfortunate <span class="code-keyword">for</span> all users, and it's name is not very descriptive either. Add d_obtain_alias as a <span class="code-keyword">new</span> exported helper that drops the inode reference in the failure <span class="code-keyword">case</span>, too and allows to pass-through NULL pointers and inodes to allow <span class="code-keyword">for</span> tail-calls in the export operations. Incidentally <span class="code-keyword">this</span> helper already existed as a <span class="code-keyword">private</span> function in libfs.c as exportfs_d_alloc so kill that one and <span class="code-keyword">switch</span> the callers to d_obtain_alias. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> diff --git a/fs/dcache.c b/fs/dcache.c index e7a1a99..46fc782 100644 --- a/fs/dcache.c +++ b/fs/dcache.c @@ -1174,6 +1174,41 @@ struct dentry * d_alloc_anon(struct inode *inode) <span class="code-keyword">return</span> res; } +/** + * d_obtain_alias - find or allocate a dentry <span class="code-keyword">for</span> a given inode + * @inode: inode to allocate the dentry <span class="code-keyword">for</span> + * + * Obtain a dentry <span class="code-keyword">for</span> an inode resulting from NFS filehandle conversion or + * similar open by handle operations. The returned dentry may be anonymous, + * or may have a full name (<span class="code-keyword">if</span> the inode was already in the cache). + * + * When called on a directory inode, we must ensure that the inode only ever + * has one dentry. If a dentry is found, that is returned instead of + * allocating a <span class="code-keyword">new</span> one. + * + * On successful <span class="code-keyword">return</span>, the reference to the inode has been transferred + * to the dentry. If %NULL is returned (indicating kmalloc failure), + * the reference on the inode has been released. To make it easier + * to use in export operations a NULL or IS_ERR inode may be passed in + * and will be casted to the corresponding NULL or IS_ERR dentry. + */ +struct dentry *d_obtain_alias(struct inode *inode) +{ + struct dentry *dentry; + + <span class="code-keyword">if</span> (!inode) + <span class="code-keyword">return</span> NULL; + <span class="code-keyword">if</span> (IS_ERR(inode)) + <span class="code-keyword">return</span> ERR_CAST(inode); + + dentry = d_alloc_anon(inode); + <span class="code-keyword">if</span> (!dentry) { + iput(inode); + dentry = ERR_PTR(-ENOMEM); + } + <span class="code-keyword">return</span> dentry; +} +EXPORT_SYMBOL_GPL(d_obtain_alias); </pre> </div></div> <p>current code is:</p> <div class="code panel" style="border-width: 1px;"><div class="codeContent panelContent"> <pre class="code-java"><span class="code-keyword">static</span> struct dentry *__d_obtain_alias(struct inode *inode, <span class="code-object">int</span> disconnected) { <span class="code-keyword">static</span> <span class="code-keyword">const</span> struct qstr anonstring = QSTR_INIT(<span class="code-quote">"/"</span>, 1); struct dentry *tmp; struct dentry *res; unsigned add_flags; <span class="code-keyword">if</span> (!inode) <span class="code-keyword">return</span> ERR_PTR(-ESTALE); <span class="code-keyword">if</span> (IS_ERR(inode)) <span class="code-keyword">return</span> ERR_CAST(inode); res = d_find_any_alias(inode); <span class="code-keyword">if</span> (res) <span class="code-keyword">goto</span> out_iput; tmp = __d_alloc(inode->i_sb, &anonstring); <span class="code-keyword">if</span> (!tmp) { res = ERR_PTR(-ENOMEM); <span class="code-keyword">goto</span> out_iput; } spin_lock(&inode->i_lock); res = __d_find_any_alias(inode); <span class="code-keyword">if</span> (res) { spin_unlock(&inode->i_lock); dput(tmp); <span class="code-keyword">goto</span> out_iput; } <span class="code-comment">/* attach a disconnected dentry */</span> add_flags = d_flags_for_inode(inode); <span class="code-keyword">if</span> (disconnected) add_flags |= DCACHE_DISCONNECTED; spin_lock(&tmp->d_lock); __d_set_inode_and_type(tmp, inode, add_flags); hlist_add_head(&tmp->d_u.d_alias, &inode->i_dentry); hlist_bl_lock(&tmp->d_sb->s_anon); hlist_bl_add_head(&tmp->d_hash, &tmp->d_sb->s_anon); hlist_bl_unlock(&tmp->d_sb->s_anon); spin_unlock(&tmp->d_lock); spin_unlock(&inode->i_lock); security_d_instantiate(tmp, inode); <span class="code-keyword">return</span> tmp; out_iput: <span class="code-keyword">if</span> (res && !IS_ERR(res)) security_d_instantiate(res, inode); iput(inode); <span class="code-keyword">return</span> res; } </pre> </div></div> <p>Oleg Drokin (oleg.drokin@intel.com) merged in patch <a href="http://review.whamcloud.com/15480/" class="external-link" target="_blank" rel="nofollow noopener">http://review.whamcloud.com/15480/</a><br/> Subject: <a href="https://jira.whamcloud.com/browse/LU-6794" title="memory leak in Lustre NFS support code, LASSERT() at unmount" class="issue-link" data-issue-key="LU-6794"><del>LU-6794</del></a> nfs: ASSERTION( atomic_read(&d->ld_ref) == 0 ) failed<br/> Project: fs/lustre-release<br/> Branch: master<br/> Current Patch Set: <br/> Commit: a37a5a9e5e309f87db9497d187df41d0fede8d29</p> <p>Landed for 2.8</p> Related LU-4272 Development Rank 1|hzxh4n: Rank (Obsolete) 9223372036854775807 Severity