linux-kernel - Re: [PATCH v5 68/68] sysfs(2): fs_index() argument is _not

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20260114165109.60e46e14@pumpkin>
Date: Wed, 14 Jan 2026 16:51:09 +0000
From: David Laight <david.laight.linux@...il.com>
To: Al Viro <viro@...iv.linux.org.uk>
Cc: linux-fsdevel@...r.kernel.org, Linus Torvalds
 <torvalds@...ux-foundation.org>, Christian Brauner <brauner@...nel.org>,
 Jan Kara <jack@...e.cz>, Mateusz Guzik <mjguzik@...il.com>, Paul Moore
 <paul@...l-moore.com>, Jens Axboe <axboe@...nel.dk>, audit@...r.kernel.org,
 io-uring@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v5 68/68] sysfs(2): fs_index() argument is _not_ a
 pathname

On Wed, 14 Jan 2026 14:35:55 +0000
Al Viro <viro@...iv.linux.org.uk> wrote:

> On Wed, Jan 14, 2026 at 10:41:55AM +0000, David Laight wrote:
> > On Wed, 14 Jan 2026 04:33:10 +0000
> > Al Viro <viro@...iv.linux.org.uk> wrote:
> >   
> > > ... it's a filesystem type name.
> > > 
> > > Signed-off-by: Al Viro <viro@...iv.linux.org.uk>
> > > ---
> > >  fs/filesystems.c | 9 +++------
> > >  1 file changed, 3 insertions(+), 6 deletions(-)
> > > 
> > > diff --git a/fs/filesystems.c b/fs/filesystems.c
> > > index 95e5256821a5..0c7d2b7ac26c 100644
> > > --- a/fs/filesystems.c
> > > +++ b/fs/filesystems.c
> > > @@ -132,24 +132,21 @@ EXPORT_SYMBOL(unregister_filesystem);
> > >  static int fs_index(const char __user * __name)
> > >  {
> > >  	struct file_system_type * tmp;
> > > -	struct filename *name;
> > > +	char *name __free(kfree) = strndup_user(__name, PATH_MAX);
> > >  	int err, index;
> > >  
> > > -	name = getname(__name);
> > > -	err = PTR_ERR(name);
> > >  	if (IS_ERR(name))
> > > -		return err;
> > > +		return PTR_ERR(name);  
> > 
> > Doesn't that end up calling kfree(name) and the check in kfree() doesn't
> > seem to exclude error values.  
> 
> include/linux/slab.h:523:DEFINE_FREE(kfree, void *, if (!IS_ERR_OR_NULL(_T)) kfree(_T))
> 
> kfree() the function won't be even called in that case...

I wasn't expecting the code to be optimised for the pointer being invalid.

I guess one of the defines does a 'dance' so that the pointer can be returned
without kfree() being called - and that needs a check in the function itself.
(I'm sure I remember something about the compiler optimising at all away.)

Perhaps the test could be:
	if (!statically_true(IS_ERR_OR_NULL(_T)) kfree(_T)
adjusting the check in kfree() to ignore -4096..16 not just 0..16.
That should reduce code size without slowing down the 'normal' paths
and possibly speeding up the error paths.

	David