Linux-Fsdevel Archive on
help / color / mirror / Atom feed
From: Amir Goldstein <>
To: Dan Carpenter <>
Cc: linux-fsdevel <>, Jan Kara <>
Subject: Re: [bug report] fsnotify: pass dir and inode arguments to fsnotify()
Date: Thu, 30 Jul 2020 14:55:11 +0300	[thread overview]
Message-ID: <> (raw)
In-Reply-To: <20200730111339.GA54272@mwanda>

On Thu, Jul 30, 2020 at 2:13 PM <> wrote:
> Hello Amir Goldstein,
> This is a semi-automatic email about new static checker warnings.
> The patch 40a100d3adc1: "fsnotify: pass dir and inode arguments to
> fsnotify()" from Jul 22, 2020, leads to the following Smatch
> complaint:

That's an odd report, because...

>     fs/notify/fsnotify.c:460 fsnotify()
>     warn: variable dereferenced before check 'inode' (see line 449)
> fs/notify/fsnotify.c
>    448          }
>    449          sb = inode->i_sb;
>                      ^^^^^^^^^^^
> New dreference.

First of all, two lines above we have
if (!inode) inode = dir;

This function does not assert (inode || dir), but must it??
This is even documented:

 * @inode:      optional inode associated with event -
 *              either @dir or @inode must be non-NULL.

The line above was indeed added by:
40a100d3adc1: "fsnotify: pass dir and inode arguments to fsnotify()"


>    450
>    451          /*
>    452           * Optimization: srcu_read_lock() has a memory barrier which can
>    453           * be expensive.  It protects walking the *_fsnotify_marks lists.
>    454           * However, if we do not walk the lists, we do not have to do
>    455           * SRCU because we have no references to any objects and do not
>    456           * need SRCU to keep them "alive".
>    457           */
>    458          if (!sb->s_fsnotify_marks &&
>    459              (!mnt || !mnt->mnt_fsnotify_marks) &&
>    460              (!inode || !inode->i_fsnotify_marks) &&
>                      ^^^^^^
> Check too late.  Presumably this check can be removed?

But this line was only added later by:
9b93f33105f5 fsnotify: send event with parent/name info to
sb/mount/non-dir marks

So, yes, the check could be removed.
It is a leftover from a previous revision, but even though it is a leftover
I kind of like the code better this way.

In principle, an event on sb/mnt that is not associated with any inode
(for example
FS_UNMOUNT) could be added in the future.
And then we will have to fix documentation and the inode dereference above.

In any case, thank you for the report, but I don't see a reason to make any
changes right now.


  reply	other threads:[~2020-07-30 11:55 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-07-30 11:13 dan.carpenter
2020-07-30 11:55 ` Amir Goldstein [this message]
2020-07-30 19:25   ` Jan Kara

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:

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

  git send-email \ \ \ \ \ \
    --subject='Re: [bug report] fsnotify: pass dir and inode arguments to fsnotify()' \

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

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).