Re: [PATCH v2 2/2] KVM: s390: Use memdup_user() rather than duplicating code

[Geert Uytterhoeven <geert@linux-m68k.org>]

Hi Markus,

On Mon, Oct 3, 2016 at 2:28 PM, SF Markus Elfring
<elfring@users.sourceforge.net> wrote:
>>> -       if (!bp_data) {
>>> -               ret = -ENOMEM;
>>> -               goto error;
>>> -       }
>>> -
>>> -       if (copy_from_user(bp_data,
>>> -                          dbg->arch.hw_bp,
>>> -                          sizeof(*bp_data) * dbg->arch.nr_hw_bp)) {
>>> -               ret = -EFAULT;
>>> -               goto error;
>>> -       }
>>> +       bp_data = memdup_user(dbg->arch.hw_bp,
>>> +                             sizeof(*bp_data) * dbg->arch.nr_hw_bp);
>>
>> ... while this would continue silently,
>
> How do you think about to explain this information a bit more?

kmalloc_array() has a builtin check for overflow while calculating the size.
This is the real reason why it's better to use kmalloc_array() than
kzalloc(n * size). If "n * size" overflow, kzalloc(n * size) would allocate a
memory block with a bogus size.

>> and corrupt memory.
>
> I wonder about this conclusion at the moment.
>
> Did you notice the check "IS_ERR(bp_data)" and the corresponding reaction
> in this update suggestion?

Yes, but bp_data may still be a valid (as in "not an error") value.

Your commit a1708a2eaded836b ("KVM: s390: Improve determination of sizes in
kvm_s390_import_bp_data()") made the code more robust, as kmalloc_array() ha
 a builtin overflow check, and will return NULL if overflow is detected.
However, commit 0624a8eb82efd58e ("KVM: s390: Use memdup_user() rather than
duplicating code") dropped that safety net again.

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds