linux-kernel - Re: [PATCH] UBIFS: Remove some dead code

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

Message-ID: <alpine.DEB.2.10.1611012022000.6300@hadrien>
Date:   Tue, 1 Nov 2016 20:28:21 +0100 (CET)
From:   Julia Lawall <julia.lawall@...6.fr>
To:     Christophe JAILLET <christophe.jaillet@...adoo.fr>
cc:     kernel-janitors@...r.kernel.org, linux-mtd@...ts.infradead.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] UBIFS: Remove some dead code



On Tue, 1 Nov 2016, Christophe JAILLET wrote:

> Le 01/11/2016 à 12:42, Richard Weinberger a écrit :
> > On 01.11.2016 07:45, Christophe JAILLET wrote:
> > > 'ubifs_fast_find_freeable()' can not return an error pointer, so this test
> > > can be removed.
> > >
> > > Signed-off-by: Christophe JAILLET <christophe.jaillet@...adoo.fr>
> > > ---
> > >   fs/ubifs/gc.c | 4 ----
> > >   1 file changed, 4 deletions(-)
> > >
> > > diff --git a/fs/ubifs/gc.c b/fs/ubifs/gc.c
> > > index e845c64b6ce1..7b35e3d6cde7 100644
> > > --- a/fs/ubifs/gc.c
> > > +++ b/fs/ubifs/gc.c
> > > @@ -846,10 +846,6 @@ int ubifs_gc_start_commit(struct ubifs_info *c)
> > >   	 */
> > >   	while (1) {
> > >   		lp = ubifs_fast_find_freeable(c);
> > > -		if (IS_ERR(lp)) {
> > > -			err = PTR_ERR(lp);
> > > -			goto out;
> > > -		}
> > Good catch, how did you find this?
> > If you have a tool/script I'd like to use it too.
> >
> > Thanks,
> > //richard
> > --
>
> Hi,
> well, it is a bit tricky.
>
> AFAIK, coccinelle is only able to match things in a given file. Finding issues
> between 2 files can be tricky.
>
> So first, I have built a list a functions which are likely to return NULL,
> either because they explicitly return NULL or if its return value is tested
> against NULL or not. See coccinelle script n°1 below.
> Then I have built a list of functions followed by a test with IS_ERR. See
> coccinelle script n°2 below.
>
> These 2 scripts generate 2 lists of functions.
> If a function is present in the 2 files, it is likely that something is
> spurious.
>
> Either the IS_ERR is not needed (this is the case in the patch above), either
> the return value is incorrectly checked. Could also be that NULL is returned
> but an error pointer would be a better option.
>
>
> I also did more or less the same for functions that return PTR_ERR and
> functions that are not followed by a test with IS_ERR.
> I can post these other scripts if wanted.
>
>
>
> Any ideas to improve or speed-up the coccinelle scripts are welcome.
> Julia ?

I made a combination of an OCaml program and a Coccinelle script to
collect the error codes (-ENOMEM, etc) that a function is returning, fully
interprocedurally throughout the kernel.  I think it ran for 17 iterations
until reaching a fixed point.  For the information I collected, it ran in
a few hours on an 8 core machine.  I think it could be repurposed to address
this NULL vs ERR_PTR problem.  I'm rolling through the Dutch countryside at
the moment, but I could take a look tomorrow.  Without going to a full
fixpoint iteration, the above strategy looks reasonable.  It could be
interesting to compare the results.  My fixpoint strategy gives up on
function pointers, so it is not completely accurate either.

julia