linux-kernel - Re: [PATCH] clk: fix a missing-free bug in clk_cpy

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]

Date:   Fri, 7 Jun 2019 09:52:58 +0800
From:   Gen Zhang <blackgod016574@...il.com>
To:     Stephen Boyd <sboyd@...nel.org>
Cc:     Jiri Slaby <jslaby@...e.cz>, mturquette@...libre.com,
        linux-clk@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] clk: fix a missing-free bug in clk_cpy_name()

On Thu, Jun 06, 2019 at 01:16:45PM -0700, Stephen Boyd wrote:
> Quoting Gen Zhang (2019-06-05 09:00:43)
> > On Wed, Jun 05, 2019 at 08:38:00AM +0200, Jiri Slaby wrote:
> > > On 31. 05. 19, 3:14, Gen Zhang wrote:
> > > > In clk_cpy_name(), '*dst_p'('parent->name'and 'parent->fw_name') and 
> > > > 'dst' are allcoted by kstrdup_const(). According to doc: "Strings 
> > > > allocated by kstrdup_const should be freed by kfree_const". So 
> > > > 'parent->name', 'parent->fw_name' and 'dst' should be freed.
> > > > 
> > > > Signed-off-by: Gen Zhang <blackgod016574@...il.com>
> > > > ---
> > > > diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
> > > > index aa51756..85c4d3f 100644
> > > > --- a/drivers/clk/clk.c
> > > > +++ b/drivers/clk/clk.c
> > > > @@ -3435,6 +3435,7 @@ static int clk_cpy_name(const char **dst_p, const char *src, bool must_exist)
> > > >     if (!dst)
> > > >             return -ENOMEM;
> > > >  
> > > > +   kfree_const(dst);
> > > 
> > > So you are now returning a freed pointer in dst_p?
> > Thanks for your reply. I re-examined the code, and this kfree is 
> > incorrect and it should be deleted.
> > > 
> > > >     return 0;
> > > >  }
> > > >  
> > > > @@ -3491,6 +3492,8 @@ static int clk_core_populate_parent_map(struct clk_core *core)
> > > >                             kfree_const(parents[i].name);
> > > >                             kfree_const(parents[i].fw_name);
> > > >                     } while (--i >= 0);
> > > > +                   kfree_const(parent->name);
> > > > +                   kfree_const(parent->fw_name);
> > > 
> > > Both of them were just freed in the loop above, no?
> > for (i = 0, parent = parents; i < num_parents; i++, parent++)
> > Is 'parent' the same as the one from the loop above?
> 
> Yes. Did it change somehow?
parent++?
> 
> > 
> > Moreover, should 'parents[i].name' and 'parents[i].fw_name' be freed by
> > kfree_const()?
> > 
> 
> Yes? They're allocated with kstrdup_const() in clk_cpy_name(), or
> they're NULL by virtue of the kcalloc and then kfree_const() does
> nothing.
I re-examined clk_cpy_name(). They are the second parameter of 
clk_cpy_name(). The first parameter is allocated, not the second one.
So 'parent->name' and 'parent->fw_name' should be freed, not 
'parents[i].name' or 'parents[i].fw_name'. Am I totally misunderstanding
this clk_cpy_name()? :-(

Thanks
Gen
> 
> I'm having a hard time following what this patch is trying to fix. It
> looks unnecessary though so I'm going to drop it from the clk review
> queue.
>