Re: [PATCH v2] ext4: fix bug in extents parsing when eh_entries == 0 and eh_depth > 0

From: Luís Henriques
Date: Fri Aug 12 2022 - 09:21:11 EST


Hi Baokun!

On Fri, Aug 12, 2022 at 08:50:34PM +0800, Baokun Li wrote:
> Hi Luís,
...
> > diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
> > index 53cfe2c681c4..a5457ac1999c 100644
> > --- a/fs/ext4/extents.c
> > +++ b/fs/ext4/extents.c
> > @@ -460,6 +460,11 @@ static int __ext4_ext_check(const char *function, unsigned int line,
> > error_msg = "invalid eh_entries";
> > goto corrupted;
> > }
> > + if (unlikely((le16_to_cpu(eh->eh_entries) == 0) &&
> > + (le16_to_cpu(eh->eh_depth > 0)))) {
>
> The parentheses are misplaced,

I'm not sure I understand what you mean. I want to have

if (unlikely((CONDITION A) && (CONDITION B))) {
/* ... */
}

so they look correct. Or is that a matter of style/alignment? (Which
checkpatch.pl doesn't complains about, by the way.)

>and le16_to_cpu is not needed here.

OK, I guess that, since both conditions do a comparison against '0', the
le16_to_cpu() can be dropped. And, if the parentheses problem you
mentioned above is a style problem, dropping it will also solve it because
that statement will become

if (unlikely((eh->eh_entries == 0) && (eh->eh_depth > 0))) {
/* ... */
}

And once again, thanks for your review!

Cheers,
--
Luís

>
> > + error_msg = "eh_entries is 0 but eh_depth is > 0";
> > + goto corrupted;
> > + }
> > if (!ext4_valid_extent_entries(inode, eh, lblk, &pblk, depth)) {
> > error_msg = "invalid extent entries";
> > goto corrupted;
> > .
>
> --
> With Best Regards,
> Baokun Li
>