Received-SPF: pass (google.com: domain of linux-ext4-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   =?utf-8?Q?Lu=C3=ADs_Henriques?= <lhenriques@suse.de>
To:     Eric Biggers <ebiggers@kernel.org>
Cc:     Theodore Ts'o <tytso@mit.edu>,
        Andreas Dilger <adilger.kernel@dilger.ca>,
        Daniel Rosenberg <drosen@google.com>,
        linux-ext4@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2] ext4: fix memory leaks in
 ext4_fname_{setup_filename,prepare_lookup}
In-Reply-To: <20230803041918.GA1218@sol.localdomain> (Eric Biggers's message
        of "Wed, 2 Aug 2023 21:19:18 -0700")
References: <20230802094931.18215-1-lhenriques@suse.de>
        <20230803041918.GA1218@sol.localdomain>
Date:   Thu, 03 Aug 2023 09:58:15 +0100
Message-ID: <87sf90v6xk.fsf@suse.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

Eric Biggers <ebiggers@kernel.org> writes:

> On Wed, Aug 02, 2023 at 10:49:31AM +0100, Lu=C3=ADs Henriques wrote:
>> If casefolding the filename fails, we'll be leaking fscrypt_buf name.
>
> fscrypt_buf =3D> fscrypt_name
>
>> diff --git a/fs/ext4/crypto.c b/fs/ext4/crypto.c
>> index e20ac0654b3f..3c05c7f3415b 100644
>> --- a/fs/ext4/crypto.c
>> +++ b/fs/ext4/crypto.c
>> @@ -33,6 +33,8 @@ int ext4_fname_setup_filename(struct inode *dir, const=
 struct qstr *iname,
>>	struct fscrypt_name name;
>>	int err;
>>
>>	err =3D fscrypt_setup_filename(dir, iname, lookup, &name);
>>	if (err)
>>		return err;
>>
>>	ext4_fname_from_fscrypt_name(fname, &name);
>>
>>  #if IS_ENABLED(CONFIG_UNICODE)
>>  	err =3D ext4_fname_setup_ci_filename(dir, iname, fname);
>> +	if (err)
>> +		fscrypt_free_filename(&name);
>>  #endif
>>  	return err;
>>  }
>> @@ -51,6 +53,8 @@ int ext4_fname_prepare_lookup(struct inode *dir, struc=
t dentry *dentry,
>>	struct fscrypt_name name;
>>	int err;
>>
>>	err =3D fscrypt_prepare_lookup(dir, dentry, &name);
>>	if (err)
>>		return err;
>>
>>	ext4_fname_from_fscrypt_name(fname, &name);
>>
>>  #if IS_ENABLED(CONFIG_UNICODE)
>>  	err =3D ext4_fname_setup_ci_filename(dir, &dentry->d_name, fname);
>> +	if (err)
>> +		fscrypt_free_filename(&name);
>>  #endif
>>  	return err;
>>  }
>
> This works, but it's a bit weird that the freeing happens on the original=
 struct
> fscrypt_name after it has already been "moved" to the struct ext4_filenam=
e by
> ext4_fname_from_fscrypt_name().  That leaves a dangling pointer in the st=
ruct
> ext4_filename.  Maybe you should call ext4_fname_free_filename() instead,=
 even
> though it would do some unnecessary work?

That makes sense, specially because fname is a parameter and it's probably
a good idea to clean-up everything before returning an error.  Thanks.

Cheers,
--=20
Lu=C3=ADs