Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.38 as permitted sender) client-ip=23.128.96.38;
Date:   Thu, 7 Dec 2023 00:03:51 -0500
From:   Kent Overstreet <kent.overstreet@linux.dev>
To:     Dave Chinner <david@fromorbit.com>
Cc:     Al Viro <viro@zeniv.linux.org.uk>, linux-fsdevel@vger.kernel.org,
        linux-block@vger.kernel.org, linux-cachefs@redhat.com,
        dhowells@redhat.com, gfs2@lists.linux.dev,
        dm-devel@lists.linux.dev, linux-security-module@vger.kernel.org,
        selinux@vger.kernel.org, linux-kernel@vger.kernel.org,
        Suren Baghdasaryan <surenb@google.com>
Subject: Re: [PATCH 03/11] vfs: Use dlock list for superblock's inode list
Message-ID: <20231207050351.gg74kg6jumik36gs@moria.home.lan>
References: <20231206060629.2827226-1-david@fromorbit.com>
 <20231206060629.2827226-4-david@fromorbit.com>
 <20231207024024.GU1674809@ZenIV>
 <ZXFRHo3mcbKfoC8v@dread.disaster.area>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <ZXFRHo3mcbKfoC8v@dread.disaster.area>
Precedence: bulk

On Thu, Dec 07, 2023 at 03:59:10PM +1100, Dave Chinner wrote:
> On Thu, Dec 07, 2023 at 02:40:24AM +0000, Al Viro wrote:
> > On Wed, Dec 06, 2023 at 05:05:32PM +1100, Dave Chinner wrote:
> > 
> > > @@ -303,6 +303,7 @@ static void destroy_unused_super(struct super_block *s)
> > >  	super_unlock_excl(s);
> > >  	list_lru_destroy(&s->s_dentry_lru);
> > >  	list_lru_destroy(&s->s_inode_lru);
> > > +	free_dlock_list_heads(&s->s_inodes);
> > >  	security_sb_free(s);
> > >  	put_user_ns(s->s_user_ns);
> > >  	kfree(s->s_subtype);
> > 
> > Umm...  Who's going to do that on normal umount?
> 
> Huh. So neither KASAN nor kmemleak has told me that s->s-inodes was
> being leaked.  I'm guessing a rebase sometime in the past silently
> dropped a critical hunk from deactivate_locked_super() in the bit
> bucket, but as nothing since whenever that happened has failed or
> flagged a memory leak I didn't notice.
> 
> Such great tools we have......

kmemleak has always seemed flakey to me (as one would expect, it's
difficult code to test). If we can ever get memory allocation profiling
merged - it won't be a direct replacement but it'll be more reliable.