linux-kernel - Re: [PATCH] mtd: rawnand: mxc: remove __init qualifier from mxcnd_probe

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:   Tue, 26 Jun 2018 09:24:13 +0200
From:   Boris Brezillon <boris.brezillon@...tlin.com>
To:     Miquel Raynal <miquel.raynal@...tlin.com>
Cc:     Martin Kaiser <martin@...ser.cx>,
        David Woodhouse <dwmw2@...radead.org>,
        Sascha Hauer <s.hauer@...gutronix.de>,
        Fabio Estevam <fabio.estevam@....com>,
        linux-mtd@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] mtd: rawnand: mxc: remove __init qualifier from
 mxcnd_probe_dt

On Tue, 26 Jun 2018 09:10:05 +0200
Miquel Raynal <miquel.raynal@...tlin.com> wrote:

> Hi Martin, Boris,
> 
> On Fri, 22 Jun 2018 16:50:25 +0200, Martin Kaiser <martin@...ser.cx>
> wrote:
> 
> > mxcnd_probe_dt is called from mxcnd_probe, which is not marked as __init.
> >   
> 
> I think this line is just a side note and should be at the end of the
> commit log.
> 
> > Using the sysfs unbind, bind nodes, mxcnd_probe and mxcnd_probe_dt can
> > potentially be called at any time. After the __init functions are cleaned,
> > mxcnd_probe_dt is no longer available. Calling it anyway causes a crash.  
> 
> The above paragraph explains the problem and the solution, you can just
> add something like "mcvnd_probe is untouched because it is already not
> marked as __init" instead of the first line.
> 
> > 
> > Signed-off-by: Martin Kaiser <martin@...ser.cx>  
> 
> Boris, do you think this patch is a good candidate for stable?
> 
> If yes, Martin, could you please add a couple of stable/fixes tags above
> your SoB?

We should at least have a Fixes tag. For the stable one, I'm not so
sure because the bug does not exist in practice (the compiler always
inline mxcnd_probe_dt()).

> 
> Thanks,
> Miquèl
> 
> > ---
> > I hope that my understanding is correct.
> > 
> > In practice, I could not get the kernel to crash, my compiler would
> > inline mxcnd_probe_dt into mxcnd_probe and ignore the __init.
> > 
> > If I marked mxcnd_probe_dt as noinline, I could trigger a crash as
> > follows
> > 
> > [root@...t ]# echo bb000000.nand > /sys/bus/platform/drivers/mxc_nand/unbind
> > [root@...t ]# echo bb000000.nand > /sys/bus/platform/drivers/mxc_nand/bind
> > Internal error: Oops - undefined instruction: 0 [#1] ARM
> > ...
> > 
> >  drivers/mtd/nand/raw/mxc_nand.c | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/mtd/nand/raw/mxc_nand.c b/drivers/mtd/nand/raw/mxc_nand.c
> > index 26cef21..90cfb5e 100644
> > --- a/drivers/mtd/nand/raw/mxc_nand.c
> > +++ b/drivers/mtd/nand/raw/mxc_nand.c
> > @@ -1686,7 +1686,7 @@ static const struct of_device_id mxcnd_dt_ids[] = {
> >  };
> >  MODULE_DEVICE_TABLE(of, mxcnd_dt_ids);
> >  
> > -static int __init mxcnd_probe_dt(struct mxc_nand_host *host)
> > +static int mxcnd_probe_dt(struct mxc_nand_host *host)
> >  {
> >  	struct device_node *np = host->dev->of_node;
> >  	const struct of_device_id *of_id =
> > @@ -1700,7 +1700,7 @@ static int __init mxcnd_probe_dt(struct mxc_nand_host *host)
> >  	return 0;
> >  }
> >  #else
> > -static int __init mxcnd_probe_dt(struct mxc_nand_host *host)
> > +static int mxcnd_probe_dt(struct mxc_nand_host *host)
> >  {
> >  	return 1;
> >  }