Re: [PATCH 4/8] ntlmv2/ntlmssp define, declare, and use crypto hash functions

Re: [PATCH 4/8] ntlmv2/ntlmssp define, declare, and use crypto hash functions

[Shirish Pargaonkar]

On Thu, Sep 9, 2010 at 7:00 AM, Suresh Jayaraman <sjayaraman@suse.de> wrote:
> On 09/08/2010 10:15 AM, shirishpargaonkar@gmail.com wrote:
>> From: Shirish Pargaonkar <shirishpargaonkar@gmail.com>
>>
>>
>> Allocate crypto hashing functions, ecurity descriptiors, and respective
>> contexts when a smb/tcp connection is established.
>> Release them when a tcp/smb connection is taken down.
>>
>> md5 and hmac-md5 are two crypto hashing functions that are used
>> throught the life of an smb/tcp connection by various functions that
>> calcualte signagure and ntlmv2 hash, HMAC etc.
>>
>>
>> Signed-off-by: Shirish Pargaonkar <shirishpargaonkar@gmail.com>
>> ---
>>  fs/cifs/cifsencrypt.c |   71 +++++++++++++++++++++++++++++++++++++++++++++++++
>>  fs/cifs/cifsproto.h   |    2 +
>>  fs/cifs/connect.c     |   16 +++++++++--
>>  3 files changed, 86 insertions(+), 3 deletions(-)
>>
>> diff --git a/fs/cifs/cifsencrypt.c b/fs/cifs/cifsencrypt.c
>> index 4bdcf13..4772c4d 100644
>> --- a/fs/cifs/cifsencrypt.c
>> +++ b/fs/cifs/cifsencrypt.c
>> @@ -369,3 +369,74 @@ void CalcNTLMv2_response(const struct cifsSesInfo *ses,
>>       hmac_md5_final(v2_session_response, &context);
>>  /*   cifs_dump_mem("v2_sess_rsp: ", v2_session_response, 32); */
>>  }
>> +
>> +void
>> +cifs_crypto_shash_release(struct TCP_Server_Info *server)
>> +{
>> +     if (server->secmech.md5)
>> +             crypto_free_shash(server->secmech.md5);
>> +
>> +     if (server->secmech.hmacmd5)
>> +             crypto_free_shash(server->secmech.hmacmd5);
>> +
>> +     kfree(server->secmech.sdeschmacmd5);
>> +
>> +     kfree(server->secmech.sdescmd5);
>> +}
>> +
>> +int
>> +cifs_crypto_shash_allocate(struct TCP_Server_Info *server)
>> +{
>> +     int rc;
>> +     unsigned int size;
>> +
>> +     server->secmech.hmacmd5 = crypto_alloc_shash("hmac(md5)", 0, 0);
>> +     if (!server->secmech.hmacmd5 ||
>> +                     IS_ERR(server->secmech.hmacmd5)) {
>
> crypto_alloc_hash() seems to return a pointer to struct crypto_shash.
> Would it be sufficient to use IS_ERR() to check?

Suresh, not sure I understand, I check for NULL value of what
crypto_alloc_shash()
returns.  IS_ERR() is what crypto code is using.
Copying crypto folks on the this thread.

>
> Also, instead of returning 1, we should use PTR_ERR() to propagate the
> appropriate error back. Otherwise, rc will always be 1 which gives
> little clue in case of error..

Done.

>
>
>> +             cERROR(1, "could not allocate crypto hmacmd5\n");
>> +             return 1;
>> +     }
>> +
>> +     server->secmech.md5 = crypto_alloc_shash("md5", 0, 0);
>> +     if (!server->secmech.md5 || IS_ERR(server->secmech.md5)) {
>> +             cERROR(1, "could not allocate crypto md5\n");
>> +             rc = 1;
>
> ditto here..

Done here too.

>
>> +             goto cifs_crypto_shash_allocate_ret1;
>
> nit: the goto labels could use better names..?

OK, will make them more meaningful in the next iteration.

>
>> +     }
>
>> +
>> +     size = sizeof(struct shash_desc) +
>> +                     crypto_shash_descsize(server->secmech.hmacmd5);
>> +     server->secmech.sdeschmacmd5 = kmalloc(size, GFP_KERNEL);
>> +     if (!server->secmech.sdeschmacmd5) {
>> +             cERROR(1, "cifs_crypto_shash_allocate: can't alloc hmacmd5\n");
>> +             rc = -ENOMEM;
>> +             goto cifs_crypto_shash_allocate_ret2;
>> +     }
>> +     server->secmech.sdeschmacmd5->shash.tfm = server->secmech.hmacmd5;
>> +     server->secmech.sdeschmacmd5->shash.flags = 0x0;
>> +
>> +
>> +     size = sizeof(struct shash_desc) +
>> +                     crypto_shash_descsize(server->secmech.md5);
>> +     server->secmech.sdescmd5 = kmalloc(size, GFP_KERNEL);
>> +     if (!server->secmech.sdescmd5) {
>> +             cERROR(1, "cifs_crypto_shash_allocate: can't alloc md5\n");
>> +             rc = -ENOMEM;
>> +             goto cifs_crypto_shash_allocate_ret3;
>> +     }
>> +     server->secmech.sdescmd5->shash.tfm = server->secmech.md5;
>> +     server->secmech.sdescmd5->shash.flags = 0x0;
>> +
>> +     return 0;
>> +
>> +cifs_crypto_shash_allocate_ret3:
>> +     kfree(server->secmech.sdeschmacmd5);
>> +
>> +cifs_crypto_shash_allocate_ret2:
>> +     crypto_free_shash(server->secmech.md5);
>> +
>> +cifs_crypto_shash_allocate_ret1:
>> +     crypto_free_shash(server->secmech.hmacmd5);
>> +
>> +     return rc;
>> +}
>> diff --git a/fs/cifs/cifsproto.h b/fs/cifs/cifsproto.h
>> index eadf78c..fa3716c 100644
>> --- a/fs/cifs/cifsproto.h
>> +++ b/fs/cifs/cifsproto.h
>> @@ -368,6 +368,8 @@ extern int cifs_calculate_mac_key(struct session_key *key, const char *rn,
>>  extern void CalcNTLMv2_response(const struct cifsSesInfo *, char *);
>>  extern void setup_ntlmv2_rsp(struct cifsSesInfo *, char *,
>>                            const struct nls_table *);
>> +extern int cifs_crypto_shash_allocate(struct TCP_Server_Info *);
>> +extern void cifs_crypto_shash_release(struct TCP_Server_Info *);
>>  #ifdef CONFIG_CIFS_WEAK_PW_HASH
>>  extern void calc_lanman_hash(const char *password, const char *cryptkey,
>>                               bool encrypt, char *lnm_session_key);
>> diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c
>> index 0ea52e9..f5369e7 100644
>> --- a/fs/cifs/connect.c
>> +++ b/fs/cifs/connect.c
>> @@ -1520,6 +1520,7 @@ cifs_put_tcp_session(struct TCP_Server_Info *server)
>>       server->tcpStatus = CifsExiting;
>>       spin_unlock(&GlobalMid_Lock);
>>
>> +     cifs_crypto_shash_release(server);
>>       cifs_fscache_release_client_cookie(server);
>>
>>       task = xchg(&server->tsk, NULL);
>> @@ -1574,10 +1575,16 @@ cifs_get_tcp_session(struct smb_vol *volume_info)
>>               goto out_err;
>>       }
>>
>> +     rc = cifs_crypto_shash_allocate(tcp_ses);
>> +     if (rc) {
>> +             cERROR(1, "could not setup hash structures rc %d", rc);
>> +             goto out_err;
>> +     }
>> +
>>       tcp_ses->hostname = extract_hostname(volume_info->UNC);
>>       if (IS_ERR(tcp_ses->hostname)) {
>>               rc = PTR_ERR(tcp_ses->hostname);
>> -             goto out_err;
>> +             goto out_err2;
>>       }
>>
>>       tcp_ses->noblocksnd = volume_info->noblocksnd;
>> @@ -1618,7 +1625,7 @@ cifs_get_tcp_session(struct smb_vol *volume_info)
>>       }
>>       if (rc < 0) {
>>               cERROR(1, "Error connecting to socket. Aborting operation");
>> -             goto out_err;
>> +             goto out_err2;
>>       }
>>
>>       /*
>> @@ -1632,7 +1639,7 @@ cifs_get_tcp_session(struct smb_vol *volume_info)
>>               rc = PTR_ERR(tcp_ses->tsk);
>>               cERROR(1, "error %d create cifsd thread", rc);
>>               module_put(THIS_MODULE);
>> -             goto out_err;
>> +             goto out_err2;
>>       }
>>
>>       /* thread spawned, put it on the list */
>> @@ -1644,6 +1651,9 @@ cifs_get_tcp_session(struct smb_vol *volume_info)
>>
>>       return tcp_ses;
>>
>> +out_err2:
>> +     cifs_crypto_shash_release(tcp_ses);
>> +
>>  out_err:
>>       if (tcp_ses) {
>>               if (!IS_ERR(tcp_ses->hostname))
>
>
> --
> Suresh Jayaraman
>

Re: [PATCH 4/8] ntlmv2/ntlmssp define, declare, and use crypto hash functions

[Suresh Jayaraman]

On 09/09/2010 09:43 PM, Shirish Pargaonkar wrote:
> On Thu, Sep 9, 2010 at 7:00 AM, Suresh Jayaraman <sjayaraman-l3A5Bk7waGM@public.gmane.org> wrote:
>> On 09/08/2010 10:15 AM, shirishpargaonkar-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org wrote:
>>> From: Shirish Pargaonkar <shirishpargaonkar-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
>>>
>>>
>>> Allocate crypto hashing functions, ecurity descriptiors, and respective
>>> contexts when a smb/tcp connection is established.
>>> Release them when a tcp/smb connection is taken down.
>>>
>>> md5 and hmac-md5 are two crypto hashing functions that are used
>>> throught the life of an smb/tcp connection by various functions that
>>> calcualte signagure and ntlmv2 hash, HMAC etc.
>>>
>>>
>>> Signed-off-by: Shirish Pargaonkar <shirishpargaonkar-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
>>> ---
>>> �fs/cifs/cifsencrypt.c | � 71 +++++++++++++++++++++++++++++++++++++++++++++++++
>>> �fs/cifs/cifsproto.h � | � �2 +
>>> �fs/cifs/connect.c � � | � 16 +++++++++--
>>> �3 files changed, 86 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/fs/cifs/cifsencrypt.c b/fs/cifs/cifsencrypt.c
>>> index 4bdcf13..4772c4d 100644
>>> --- a/fs/cifs/cifsencrypt.c
>>> +++ b/fs/cifs/cifsencrypt.c
>>> @@ -369,3 +369,74 @@ void CalcNTLMv2_response(const struct cifsSesInfo *ses,
>>> � � � hmac_md5_final(v2_session_response, &context);
>>> �/* � cifs_dump_mem("v2_sess_rsp: ", v2_session_response, 32); */
>>> �}
>>> +
>>> +void
>>> +cifs_crypto_shash_release(struct TCP_Server_Info *server)
>>> +{
>>> + � � if (server->secmech.md5)
>>> + � � � � � � crypto_free_shash(server->secmech.md5);
>>> +
>>> + � � if (server->secmech.hmacmd5)
>>> + � � � � � � crypto_free_shash(server->secmech.hmacmd5);
>>> +
>>> + � � kfree(server->secmech.sdeschmacmd5);
>>> +
>>> + � � kfree(server->secmech.sdescmd5);
>>> +}
>>> +
>>> +int
>>> +cifs_crypto_shash_allocate(struct TCP_Server_Info *server)
>>> +{
>>> + � � int rc;
>>> + � � unsigned int size;
>>> +
>>> + � � server->secmech.hmacmd5 = crypto_alloc_shash("hmac(md5)", 0, 0);
>>> + � � if (!server->secmech.hmacmd5 ||
>>> + � � � � � � � � � � IS_ERR(server->secmech.hmacmd5)) {
>>
>> crypto_alloc_hash() seems to return a pointer to struct crypto_shash.
>> Would it be sufficient to use IS_ERR() to check?
> 
> Suresh, not sure I understand, I check for NULL value of what
> crypto_alloc_shash()
> returns.  IS_ERR() is what crypto code is using.
> Copying crypto folks on the this thread.

I was suggesting to remove the NULL value check as checking for
IS_ERR(server->secmech.hmacmd5 alone can handle NULL cases and is
sufficient.



-- 
Suresh Jayaraman