public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: "Luck, Tony" <tony.luck@intel.com>
To: Al Viro <viro@ZenIV.linux.org.uk>
Cc: Xiaotian Feng <xtfeng@gmail.com>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org
Subject: pstore: fix leaking ->i_private
Date: Fri, 18 Mar 2011 11:44:48 -0700	[thread overview]
Message-ID: <4d83a82032021e2997@agluck-desktop.sc.intel.com> (raw)
In-Reply-To: <20110317104434.GB22723@ZenIV.linux.org.uk>

From: Tony Luck <tony.luck@intel.com>

Move kfree() of i_private out of ->unlink() and into ->evict_inode()

Signed-off-by: Tony Luck <tony.luck@intel.com>

---

> Incidentally, you are leaking ->i_private on umount.  You want ->evict_inode()
> doing that kfree(), not ->unlink().  And possibly ->erase() as well, with
> check for zero i_nlink around it (in ->evict_inode()).

I think this does what I want.

I only want to ->erase() the persistant store if there has been an unlink(2)
on the file.  It looks to be safe and right to do this in ->unlink()

I don't think I need any checks on i_nlink - pstore filesystem won't let you
make extra links to a file.

end_writeback(inode) looks to be somewhat overkill, all I need is to set I_CLEAR
bit in ->i_state. But nobody else does this by hand, and the extra actions in
end_writeback() look to be harmless.

diff --git a/fs/pstore/inode.c b/fs/pstore/inode.c
index 0834223..e8b0c08 100644
--- a/fs/pstore/inode.c
+++ b/fs/pstore/inode.c
@@ -73,11 +73,16 @@ static int pstore_unlink(struct inode *dir, struct dentry *dentry)
 	struct pstore_private *p = dentry->d_inode->i_private;
 
 	p->erase(p->id);
-	kfree(p);
 
 	return simple_unlink(dir, dentry);
 }
 
+static void pstore_evict_inode(struct inode *inode)
+{
+	kfree(inode->i_private);
+	end_writeback(inode);
+}
+
 static const struct inode_operations pstore_dir_inode_operations = {
 	.lookup		= simple_lookup,
 	.unlink		= pstore_unlink,
@@ -110,6 +115,7 @@ static struct inode *pstore_get_inode(struct super_block *sb,
 static const struct super_operations pstore_ops = {
 	.statfs		= simple_statfs,
 	.drop_inode	= generic_delete_inode,
+	.evict_inode	= pstore_evict_inode,
 	.show_options	= generic_show_options,
 };
 

  parent reply	other threads:[~2011-03-18 18:44 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-03-16 23:56 [git pull] mnt_devname queue Al Viro
2011-03-17  5:45 ` Xiaotian Feng
2011-03-17  7:23   ` Al Viro
2011-03-17  7:28     ` Xiaotian Feng
2011-03-17 10:44       ` Al Viro
2011-03-17 19:08         ` Tony Luck
2011-03-17 21:35         ` Some fixes for pstore (Was Re: [git pull] mnt_devname queue) Tony Luck
2011-03-17 22:42           ` Al Viro
2011-03-17 22:48             ` Tony Luck
2011-03-17 22:56               ` Al Viro
2011-03-18 18:44         ` Luck, Tony [this message]
2011-03-18 18:49           ` pstore: fix leaking ->i_private Christoph Hellwig
2011-03-18 18:55             ` Tony Luck
2011-03-18 22:33             ` pstore: use mount option instead sysfs to tweak kmsg_bytes Luck, Tony
2011-03-18 18:57           ` pstore: fix leaking ->i_private Al Viro
2011-03-17 23:29 ` Some fixes for pstore (Was Re: [git pull] mnt_devname queue) Tony Luck
2011-03-18  0:07   ` Al Viro

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=4d83a82032021e2997@agluck-desktop.sc.intel.com \
    --to=tony.luck@intel.com \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=torvalds@linux-foundation.org \
    --cc=viro@ZenIV.linux.org.uk \
    --cc=xtfeng@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox