Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20230801112337.1856215-1-zhangshida@kylinos.cn> <20230801151828.GB11332@frogsfrogsfrogs>
In-Reply-To: <20230801151828.GB11332@frogsfrogsfrogs>
From:   Stephen Zhang <starzhangzsd@gmail.com>
Date:   Wed, 2 Aug 2023 09:17:05 +0800
Message-ID: <CANubcdX2-T1QpDRmvddL-9bme4ZeFvre4N_nsw=gursx9zSQhg@mail.gmail.com>
Subject: Re: [PATCH v3] ext4: Fix rec_len verify error
To:     "Darrick J. Wong" <djwong@kernel.org>
Cc:     tytso@mit.edu, adilger.kernel@dilger.ca, yi.zhang@huawei.com,
        linux-ext4@vger.kernel.org, linux-kernel@vger.kernel.org,
        zhangshida@kylinos.cn, stable@kernel.org,
        Andreas Dilger <adilger@dilger.ca>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

Darrick J. Wong <djwong@kernel.org> =E4=BA=8E2023=E5=B9=B48=E6=9C=881=E6=97=
=A5=E5=91=A8=E4=BA=8C 23:18=E5=86=99=E9=81=93=EF=BC=9A
>
>
> This is sitll missing some pieces; what about this clause at line 367:
>
>         if (t->det_reserved_zero1 ||
>             le16_to_cpu(t->det_rec_len) !=3D sizeof(struct ext4_dir_entry=
_tail) ||
>             t->det_reserved_zero2 ||
>             t->det_reserved_ft !=3D EXT4_FT_DIR_CSUM)
>                 return NULL;
>

Yeah...

> > @@ -445,13 +445,13 @@ static struct dx_countlimit *get_dx_countlimit(st=
ruct inode *inode,
> >       struct ext4_dir_entry *dp;
> >       struct dx_root_info *root;
> >       int count_offset;
> > +     int blocksize =3D EXT4_BLOCK_SIZE(inode->i_sb);
> >
> > -     if (le16_to_cpu(dirent->rec_len) =3D=3D EXT4_BLOCK_SIZE(inode->i_=
sb))
> > +     if (ext4_rec_len_from_disk(dirent->rec_len, blocksize) =3D=3D blo=
cksize)
> >               count_offset =3D 8;
> > -     else if (le16_to_cpu(dirent->rec_len) =3D=3D 12) {
> > +     else if (ext4_rec_len_from_disk(dirent->rec_len, blocksize) =3D=
=3D 12) {
>
> Why not lift this ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ to a
> local variable?  @dirent doesn't change, right?
>

Will do.

> >               dp =3D (struct ext4_dir_entry *)(((void *)dirent) + 12);
> > -             if (le16_to_cpu(dp->rec_len) !=3D
> > -                 EXT4_BLOCK_SIZE(inode->i_sb) - 12)
> > +             if (ext4_rec_len_from_disk(dp->rec_len, blocksize) !=3D b=
locksize - 12)
> >                       return NULL;
> >               root =3D (struct dx_root_info *)(((void *)dp + 12));
> >               if (root->reserved_zero ||
>
> What about dx_make_map?
>

dx_make_map(
     ....
     map_tail->size =3D le16_to_cpu(de->rec_len);

That might be a questionable one...
map_tail->size is 16 bit, while the key reason we want ext4_rec_len_from_di=
sk
is converting 16-bit rec_len disk form to a in-memory form consisting of li=
ke 17
bits...i.e. 0x10000.

Cheers,
Shida


> Here's all the opencoded access I could find:
>
> $ git grep le16.*rec_len fs/ext4/
> fs/ext4/namei.c:356:                le16_to_cpu(d->rec_len));
> fs/ext4/namei.c:367:        le16_to_cpu(t->det_rec_len) !=3D sizeof(struc=
t ext4_dir_entry_tail) ||
> fs/ext4/namei.c:449:    if (le16_to_cpu(dirent->rec_len) =3D=3D EXT4_BLOC=
K_SIZE(inode->i_sb))
> fs/ext4/namei.c:451:    else if (le16_to_cpu(dirent->rec_len) =3D=3D 12) =
{
> fs/ext4/namei.c:453:            if (le16_to_cpu(dp->rec_len) !=3D
> fs/ext4/namei.c:1338:                   map_tail->size =3D le16_to_cpu(de=
->rec_len);
>
> --D
>
> > --
> > 2.27.0
> >