Re: [PATCH] profiling: fix shift-out-of-bounds in profile_setup

[Pavel Skripkin]

On 8/8/21 2:21 PM, Tetsuo Handa wrote:
> On 2021/07/17 4:04, Pavel Skripkin wrote:
> > Syzbot reported shift-out-of-bounds bug in profile_init().
> > The problem was in incorrect prof_shift. Since prof_shift value comes from
> > userspace we need to check this value to avoid too big shift.
> >
> > Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
> > Reported-and-tested-by: syzbot+e68c89a9510c159d9684@xxxxxxxxxxxxxxxxxxxxxxxxx
> > Suggested-by: Tetsuo Handa <penguin-kernel@xxxxxxxxxxxxxxxxxxx>
> > Signed-off-by: Pavel Skripkin <paskripkin@xxxxxxxxx>
> > ---
> >  kernel/profile.c | 9 +++++----
> >  1 file changed, 5 insertions(+), 4 deletions(-)
> >
> > diff --git a/kernel/profile.c b/kernel/profile.c
> > index c2ebddb5e974..c905931e3c3b 100644
> > --- a/kernel/profile.c
> > +++ b/kernel/profile.c
> > @@ -42,6 +42,7 @@ struct profile_hit {
> >  
> >  static atomic_t *prof_buffer;
> >  static unsigned long prof_len, prof_shift;
> > +#define MAX_PROF_SHIFT		(sizeof(prof_shift) * 8)
>
> I came to think that we should directly use BITS_PER_LONG, for
> the integer value which is subjected to shift operation is e.g.
>
> 	(_etext - _stext)
>
> part of
>
> 	prof_len = (_etext - _stext) >> prof_shift;
>
> in profile_init().
>
> Since "unsigned char" will be sufficient for holding BITS_PER_LONG - 1,
> defining MAX_PROF_SHIFT based on size of prof_shift is incorrect.

I don't get it, sorry. Do you mean, that

#define MAX_PROF_SHIFT		BITS_PER_LONG

is better solution here? And as I understand we can change prof_shift 
type from "unsigned long" to "unsigned short", rigth?



> Also, there is
>
> 	unsigned int sample_step = 1 << prof_shift;
>
> in read_profile(). This may result in shift-out-of-bounds on BITS_PER_LONG == 64
> architecture. Shouldn't this variable changed from "unsigned int" to "unsigned long"
> and use 1UL instead of 1 ?

Yep, it's necessary. Will do it in v2


With regards,
Pavel Skripkin