Re: [GIT PULL] s390 patches for 2.6.39-rc4

From: Martin Schwidefsky
Date: Wed Apr 27 2011 - 02:37:48 EST


On Tue, 26 Apr 2011 11:38:22 -0700
Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:

> I'll pull this, but:
>
> On Tue, Apr 26, 2011 at 2:03 AM, Martin Schwidefsky
> <schwidefsky@xxxxxxxxxx> wrote:
> >
> > Jan Glauber (3):
> > Â Â Â[S390] prng: prevent access beyond end of stack
>
> I don't think that's _still_ correct.
>
> Just looking at the diff, it looks broken:
>
> > Â Â Â Â/* Add the entropy */
> > Â Â Â Âwhile (nbytes >= 8) {
> > - Â Â Â Â Â Â Â *((__u64 *)parm_block) ^= *((__u64 *)buf+i*8);
> > + Â Â Â Â Â Â Â *((__u64 *)parm_block) ^= *((__u64 *)buf+i);
> > Â Â Â Â Â Â Â Âprng_add_entropy();
> > Â Â Â Â Â Â Â Âi += 8;
> > Â Â Â Â Â Â Â Ânbytes -= 8;
>
> The "i += 8" implies that "i" is a byte offset.
>
> But "(__u64 *)buf+i" will parse as "((__u64 *)buf) + i" and thus still
> do another multiply-by-eight.
>
> So the diff fixes one "off by a factor of 8", but not another, if I
> read it right.
>
> So the parenthesis should be moved around to something like "*(__u64
> *) (buf+i)", I think.

Urgs, pointer arithmetic for dummies.. it is indeed still wrong. We'll fix.

--
blue skies,
Martin.

"Reality continues to ruin my life." - Calvin.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/