linux-fsdevel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Baokun Li <libaokun@huaweicloud.com>
To: David Howells <dhowells@redhat.com>
Cc: Markus Elfring <Markus.Elfring@web.de>,
	netfs@lists.linux.dev, linux-fsdevel@vger.kernel.org,
	linux-erofs@lists.ozlabs.org, Jeff Layton <jlayton@kernel.org>,
	stable@kernel.org, LKML <linux-kernel@vger.kernel.org>,
	Christian Brauner <brauner@kernel.org>,
	Gao Xiang <hsiangkao@linux.alibaba.com>,
	Hou Tao <houtao1@huawei.com>,
	Jingbo Xu <jefflexu@linux.alibaba.com>,
	Yang Erkun <yangerkun@huawei.com>, Yu Kuai <yukuai3@huawei.com>,
	Zizhi Wo <wozizhi@huawei.com>, Baokun Li <libaokun1@huawei.com>,
	Baokun Li <libaokun@huaweicloud.com>
Subject: Re: [PATCH] cachefiles: fix dentry leak in cachefiles_open_file()
Date: Wed, 28 Aug 2024 22:05:08 +0800	[thread overview]
Message-ID: <11c591fd-221b-4eeb-a0bd-e9e303d391a6@huaweicloud.com> (raw)
In-Reply-To: <988772.1724850113@warthog.procyon.org.uk>

Hello David,

Thanks for the review.

On 2024/8/28 21:01, David Howells wrote:
> Baokun Li <libaokun@huaweicloud.com> wrote:
>
>> Actually, at first I was going to release the reference count of the
>> dentry uniformly in cachefiles_look_up_object() and delete all dput()
>> in cachefiles_open_file(),
> You couldn't do that anyway, since kernel_file_open() steals the caller's ref
> if successful.
Ignoring kernel_file_open(), we now put a reference count of the dentry
whether cachefiles_open_file() returns true or false.

And cachefiles_open_file() doesn't modify the dentry, so I'm thinking it's
releasing the reference count of the dentry that was got by
lookup_positive_unlocked() in cachefiles_look_up_object().

I'm not sure how kernel_file_open() steals the reference count,
am I missing something?


The code is as follows:

diff --git a/fs/cachefiles/namei.c b/fs/cachefiles/namei.c
index f53977169db4..2b3f9935dbb4 100644
--- a/fs/cachefiles/namei.c
+++ b/fs/cachefiles/namei.c
@@ -595,14 +595,12 @@ static bool cachefiles_open_file(struct 
cachefiles_object *object,
          * write and readdir but not lookup or open).
          */
         touch_atime(&file->f_path);
-       dput(dentry);
         return true;

  check_failed:
         fscache_cookie_lookup_negative(object->cookie);
         cachefiles_unmark_inode_in_use(object, file);
         fput(file);
-       dput(dentry);
         if (ret == -ESTALE)
                 return cachefiles_create_file(object);
         return false;
@@ -611,7 +609,6 @@ static bool cachefiles_open_file(struct 
cachefiles_object *object,
         fput(file);
  error:
         cachefiles_do_unmark_inode_in_use(object, d_inode(dentry));
-       dput(dentry);
         return false;
  }

@@ -654,7 +651,9 @@ bool cachefiles_look_up_object(struct 
cachefiles_object *object)
                 goto new_file;
         }

-       if (!cachefiles_open_file(object, dentry))
+       ret = cachefiles_open_file(object, dentry);
+       dput(dentry);
+       if (!ret)
                 return false;

         _leave(" = t [%lu]", file_inode(object->file)->i_ino);


Regards,
Baokun

>> but this may conflict when backporting the code to stable. So just keep it
>> simple to facilitate backporting to stable.
> Prioritise upstream, please.
>
> I think Markus's suggestion of inserting a label and switching to a goto is
> better.
>
> Thanks,
> David


  reply	other threads:[~2024-08-28 14:05 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-08-26  4:00 [PATCH] cachefiles: fix dentry leak in cachefiles_open_file() libaokun
2024-08-26 13:55 ` Markus Elfring
2024-08-27  3:47   ` Baokun Li
2024-08-27  8:34     ` Markus Elfring
2024-08-28 13:01   ` [PATCH] " David Howells
2024-08-28 14:05     ` Baokun Li [this message]
2024-08-28 16:14     ` David Howells
2024-08-29  1:43       ` Baokun Li

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=11c591fd-221b-4eeb-a0bd-e9e303d391a6@huaweicloud.com \
    --to=libaokun@huaweicloud.com \
    --cc=Markus.Elfring@web.de \
    --cc=brauner@kernel.org \
    --cc=dhowells@redhat.com \
    --cc=houtao1@huawei.com \
    --cc=hsiangkao@linux.alibaba.com \
    --cc=jefflexu@linux.alibaba.com \
    --cc=jlayton@kernel.org \
    --cc=libaokun1@huawei.com \
    --cc=linux-erofs@lists.ozlabs.org \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=netfs@lists.linux.dev \
    --cc=stable@kernel.org \
    --cc=wozizhi@huawei.com \
    --cc=yangerkun@huawei.com \
    --cc=yukuai3@huawei.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;
as well as URLs for NNTP newsgroup(s).