Re: [PATCH v7 5/5] tpm: pass an array of tpm_extend_digest structures to tpm_pcr_extend()

[Roberto Sassu <roberto.sassu@huawei.com>]

On 1/18/2019 4:12 PM, Jarkko Sakkinen wrote:
> On Thu, Jan 17, 2019 at 08:59:00AM +0100, Roberto Sassu wrote:
>> On 12/20/2018 4:21 PM, Jarkko Sakkinen wrote:
>>> On Thu, Dec 13, 2018 at 11:29:45AM +0100, Roberto Sassu wrote:
>>>> Currently, tpm_pcr_extend() accepts as an input only a SHA1 digest.
>>>>
>>>> This patch modifies the definition of tpm_pcr_extend() to allow other
>>>> kernel subsystems to pass a digest for each algorithm supported by the TPM.
>>>> All digests are processed by the TPM in one operation.
>>>>
>>>> If a tpm_pcr_extend() caller provides a subset of the supported algorithms,
>>>> the TPM driver extends the remaining PCR banks with the first digest
>>>> passed as an argument to the function.
>>>>
>>>> The new tpm_extend digest structure has been preferred to the tpm_digest
>>>> structure, to let the caller specify the size of the digest (which may be
>>>> unknown to the TPM driver).
>>>>
>>>> Due to the API change, ima_pcr_extend() and pcrlock() have been modified.
>>>>
>>>> Signed-off-by: Roberto Sassu <roberto.sassu@huawei.com>
>>>> ---
>>>>    drivers/char/tpm/tpm-interface.c   | 24 +++++---------------
>>>>    drivers/char/tpm/tpm.h             |  5 +++--
>>>>    drivers/char/tpm/tpm1-cmd.c        | 13 ++++++++---
>>>>    drivers/char/tpm/tpm2-cmd.c        | 35 +++++++++++++++++++++---------
>>>>    include/linux/tpm.h                | 13 ++++++++---
>>>>    security/integrity/ima/ima_queue.c |  5 ++++-
>>>>    security/keys/trusted.c            |  5 ++++-
>>>>    7 files changed, 62 insertions(+), 38 deletions(-)
>>>>
>>>> diff --git a/drivers/char/tpm/tpm-interface.c b/drivers/char/tpm/tpm-interface.c
>>>> index eb7c79ca8a94..911fea19e408 100644
>>>> --- a/drivers/char/tpm/tpm-interface.c
>>>> +++ b/drivers/char/tpm/tpm-interface.c
>>>> @@ -478,42 +478,30 @@ EXPORT_SYMBOL_GPL(tpm_pcr_read);
>>>>     * tpm_pcr_extend - extend a PCR value in SHA1 bank.
>>>>     * @chip:	a &struct tpm_chip instance, %NULL for the default chip
>>>>     * @pcr_idx:	the PCR to be retrieved
>>>> - * @hash:	the hash value used to extend the PCR value
>>>> + * @count:	number of tpm_extend_digest structures
>>>> + * @digests:	array of tpm_extend_digest structures used to extend PCRs
>>>>     *
>>>>     * Note: with TPM 2.0 extends also those banks for which no digest was
>>>>     * specified in order to prevent malicious use of those PCR banks.
>>>>     *
>>>>     * Return: same as with tpm_transmit_cmd()
>>>>     */
>>>> -int tpm_pcr_extend(struct tpm_chip *chip, u32 pcr_idx, const u8 *hash)
>>>> +int tpm_pcr_extend(struct tpm_chip *chip, u32 pcr_idx, u32 count,
>>>> +		   const struct tpm_extend_digest *digests)
>>>
>>> Remove const. Document how @digests is used  like the special meaning
>>> of the first index. I faintly remember asking this last time.
>>>
>>>>    {
>>>>    	int rc;
>>>> -	struct tpm_digest *digest_list;
>>>> -	int i;
>>>>    	chip = tpm_find_get_ops(chip);
>>>>    	if (!chip)
>>>>    		return -ENODEV;
>>>>    	if (chip->flags & TPM_CHIP_FLAG_TPM2) {
>>>> -		digest_list = kcalloc(chip->nr_allocated_banks,
>>>> -				      sizeof(*digest_list), GFP_KERNEL);
>>>> -		if (!digest_list)
>>>> -			return -ENOMEM;
>>>> -
>>>> -		for (i = 0; i < chip->nr_allocated_banks; i++) {
>>>> -			digest_list[i].alg_id = chip->allocated_banks[i].alg_id;
>>>> -			memcpy(digest_list[i].digest, hash, TPM_DIGEST_SIZE);
>>>> -		}
>>>> -
>>>> -		rc = tpm2_pcr_extend(chip, pcr_idx, chip->nr_allocated_banks,
>>>> -				     digest_list);
>>>> -		kfree(digest_list);
>>>> +		rc = tpm2_pcr_extend(chip, pcr_idx, count, digests);
>>>>    		tpm_put_ops(chip);
>>>>    		return rc;
>>>>    	}
>>>> -	rc = tpm1_pcr_extend(chip, pcr_idx, hash,
>>>> +	rc = tpm1_pcr_extend(chip, pcr_idx, count, digests,
>>>>    			     "attempting extend a PCR value");
>>>
>>> The validation is missing that the provided array has only one element
>>> and the algorithm is SHA1. Could be done also inside tpm1_pcr_extend()
>>> but what you are doing to that function does not make any sense so
>>> better to do it here.
>>>
>>>>    	tpm_put_ops(chip);
>>>>    	return rc;
>>>> diff --git a/drivers/char/tpm/tpm.h b/drivers/char/tpm/tpm.h
>>>> index 64d93d26087f..6b446504d2fe 100644
>>>> --- a/drivers/char/tpm/tpm.h
>>>> +++ b/drivers/char/tpm/tpm.h
>>>> @@ -504,7 +504,8 @@ int tpm1_auto_startup(struct tpm_chip *chip);
>>>>    int tpm1_do_selftest(struct tpm_chip *chip);
>>>>    int tpm1_get_timeouts(struct tpm_chip *chip);
>>>>    unsigned long tpm1_calc_ordinal_duration(struct tpm_chip *chip, u32 ordinal);
>>>> -int tpm1_pcr_extend(struct tpm_chip *chip, u32 pcr_idx, const u8 *hash,
>>>> +int tpm1_pcr_extend(struct tpm_chip *chip, u32 pcr_idx, u32 count,
>>>> +		    const struct tpm_extend_digest *digests,
>>>>    		    const char *log_msg);
>>>>    int tpm1_pcr_read(struct tpm_chip *chip, u32 pcr_idx, u8 *res_buf);
>>>>    ssize_t tpm1_getcap(struct tpm_chip *chip, u32 subcap_id, cap_t *cap,
>>>> @@ -551,7 +552,7 @@ int tpm2_get_timeouts(struct tpm_chip *chip);
>>>>    int tpm2_pcr_read(struct tpm_chip *chip, u32 pcr_idx,
>>>>    		  struct tpm_digest *digest, u16 *digest_size_ptr);
>>>>    int tpm2_pcr_extend(struct tpm_chip *chip, u32 pcr_idx, u32 count,
>>>> -		    struct tpm_digest *digests);
>>>> +		    const struct tpm_extend_digest *digests);
>>>>    int tpm2_get_random(struct tpm_chip *chip, u8 *dest, size_t max);
>>>>    void tpm2_flush_context_cmd(struct tpm_chip *chip, u32 handle,
>>>>    			    unsigned int flags);
>>>> diff --git a/drivers/char/tpm/tpm1-cmd.c b/drivers/char/tpm/tpm1-cmd.c
>>>> index 8b70a7f884a7..04ee10284b8c 100644
>>>> --- a/drivers/char/tpm/tpm1-cmd.c
>>>> +++ b/drivers/char/tpm/tpm1-cmd.c
>>>> @@ -449,12 +449,20 @@ int tpm1_get_timeouts(struct tpm_chip *chip)
>>>>    }
>>>>    #define TPM_ORD_PCR_EXTEND 20
>>>> -int tpm1_pcr_extend(struct tpm_chip *chip, u32 pcr_idx, const u8 *hash,
>>>> +int tpm1_pcr_extend(struct tpm_chip *chip, u32 pcr_idx, u32 count,
>>>> +		    const struct tpm_extend_digest *digests,
>>>>    		    const char *log_msg)
>>>>    {
>>>>    	struct tpm_buf buf;
>>>> +	u8 dummy_hash[TPM_DIGEST_SIZE] = { 0 };
>>>> +	const u8 *hash;
>>>>    	int rc;
>>>> +	hash = dummy_hash;
>>>> +	if (count)
>>>> +		memcpy(dummy_hash, digests[0].data,
>>>> +		       min(digests[0].size, (u16)sizeof(dummy_hash)));
>>>> +
>>>
>>> You copy memory from one place to another without any good reason to do
>>> so. My suggestion is just not to change tpm1_pcr_extend() at all.
>>>
>>>>    	rc = tpm_buf_init(&buf, TPM_TAG_RQU_COMMAND, TPM_ORD_PCR_EXTEND);
>>>>    	if (rc)
>>>>    		return rc;
>>>> @@ -743,7 +751,6 @@ int tpm1_auto_startup(struct tpm_chip *chip)
>>>>     */
>>>>    int tpm1_pm_suspend(struct tpm_chip *chip, u32 tpm_suspend_pcr)
>>>>    {
>>>> -	u8 dummy_hash[TPM_DIGEST_SIZE] = { 0 };
>>>>    	struct tpm_buf buf;
>>>>    	unsigned int try;
>>>>    	int rc;
>>>> @@ -751,7 +758,7 @@ int tpm1_pm_suspend(struct tpm_chip *chip, u32 tpm_suspend_pcr)
>>>>    	/* for buggy tpm, flush pcrs with extend to selected dummy */
>>>>    	if (tpm_suspend_pcr)
>>>> -		rc = tpm1_pcr_extend(chip, tpm_suspend_pcr, dummy_hash,
>>>> +		rc = tpm1_pcr_extend(chip, tpm_suspend_pcr, 0, NULL,
>>>>    				     "extending dummy pcr before suspend");
>>>>    	rc = tpm_buf_init(&buf, TPM_TAG_RQU_COMMAND, TPM_ORD_SAVESTATE);
>>>> diff --git a/drivers/char/tpm/tpm2-cmd.c b/drivers/char/tpm/tpm2-cmd.c
>>>> index 6ce5173cf0e5..77b5808270c6 100644
>>>> --- a/drivers/char/tpm/tpm2-cmd.c
>>>> +++ b/drivers/char/tpm/tpm2-cmd.c
>>>> @@ -247,21 +247,22 @@ struct tpm2_null_auth_area {
>>>>     *
>>>>     * @chip:	TPM chip to use.
>>>>     * @pcr_idx:	index of the PCR.
>>>> - * @count:	number of digests passed.
>>>> - * @digests:	list of pcr banks and corresponding digest values to extend.
>>>> + * @count:	number of tpm_extend_digest passed.
>>>> + * @digests:	array of tpm_extend_digest with digest values to extend.
>>>>     *
>>>>     * Return: Same as with tpm_transmit_cmd.
>>>>     */
>>>
>>> The documentation about @digests.
>>>
>>>>    int tpm2_pcr_extend(struct tpm_chip *chip, u32 pcr_idx, u32 count,
>>>> -		    struct tpm_digest *digests)
>>>> +		    const struct tpm_extend_digest *digests)
>>>>    {
>>>>    	struct tpm_buf buf;
>>>>    	struct tpm2_null_auth_area auth_area;
>>>> +	const struct tpm_extend_digest *digest;
>>>> +	u8 dummy_hash[SHA512_DIGEST_SIZE] = { 0 };
>>>> +	const u8 *hash;
>>>>    	int rc;
>>>>    	int i;
>>>> -
>>>> -	if (count > chip->nr_allocated_banks)
>>>> -		return -EINVAL;
>>>> +	int j;
>>>>    	rc = tpm_buf_init(&buf, TPM2_ST_SESSIONS, TPM2_CC_PCR_EXTEND);
>>>>    	if (rc)
>>>> @@ -277,11 +278,25 @@ int tpm2_pcr_extend(struct tpm_chip *chip, u32 pcr_idx, u32 count,
>>>>    	tpm_buf_append_u32(&buf, sizeof(struct tpm2_null_auth_area));
>>>>    	tpm_buf_append(&buf, (const unsigned char *)&auth_area,
>>>>    		       sizeof(auth_area));
>>>> -	tpm_buf_append_u32(&buf, count);
>>>> +	tpm_buf_append_u32(&buf, chip->nr_allocated_banks);
>>>> +
>>>> +	if (count)
>>>> +		memcpy(dummy_hash, digests[0].data, digests[0].size);
>>>> +
>>>> +	for (i = 0; i < chip->nr_allocated_banks; i++) {
>>>> +		tpm_buf_append_u16(&buf, chip->allocated_banks[i].alg_id);
>>>> +
>>>> +		hash = dummy_hash;
>>>> +		for (j = 0; j < count; j++) {
>>>> +			digest = digests + j;
>>>> +
>>>> +			if (digest->alg_id == chip->allocated_banks[i].alg_id) {
>>>> +				hash = digest->data;
>>>> +				break;
>>>
>>> I think the whole design is just wrong. I did re-read your response to
>>> v6 again and I'm very sorry, but I just don't get this. Caller has all
>>> the information (from struct tpm_chip) to give the correct data. This
>>> function should validate that data (check algorithm ID and that's it).
>>
>> The question is if checking tpm->allocated_banks is a strict
>> requirement, or we can allow callers to use the algorithm they are
>> currently using, without further modifications.
> 
> If you want to know what is available, you can use that array.
> 
>>> Extending with the dummy hash should be done by the caller when
>>> preparing the array, not baked into this function. This kind of also
>>> makes obvious that we don't need this new struct. There should never be
>>> a local variable (whose size is BTW randomly chosen) called dummy_hash.
>>
>> This means duplicating the code for each caller. Currently, this work is
>> done by the TPM driver.
> 
> It is better than fix the usage pattern like this. Not only that, but
> this somewhat complicated behavior now completely undocumented in the
> function description.
> 
> You are making strong assumptions how some other subsystem might use
> extend operation. I will rather accept redundancy and consider
> consolidation later when there are two clients. Rather keep the TPM
> provided function simple and stupid.

I just found that your requirement does not match Mimi's requirements:

---
 > The caller of tpm_pcr_extend() could pass a subset of the allocated
 > banks, but the TPM driver extends all banks as before.

Agreed, there should be a clear division.

1) The caller shouldn't need to know anything about the chip->info.
2) The TPM driver should not rely on the caller to supply all the
hashes, but verify that all allocated banks are being extended.
---

See https://lkml.org/lkml/2018/11/19/603.


Also, if callers should be able to retrieve the list of supported
algorithms from the TPM, I should move the tpm_chip definition to
include/linux/tpm.h or introduce a new function that returns them.

However, you said that this shouldn't be done now:

---
 > > > @@ -1048,16 +1047,22 @@ int tpm_pcr_extend(struct tpm_chip *chip, 
int pcr_idx, const u8 *hash)
 > > >   		return -ENODEV;
 > >
 > > Should take digest_list as input. Probably callers could re-use the
 > > same digest_list array multiple times?
 > >
 > > Move struct tpm_chip to include/linux/tpm.h so that the caller can 
query
 > > nr_active_banks and active_banks and can create the digest array by
 > > itself. Lets do this right at once now that this is being restructured.
 >
 > I have to move also other structures and #define. Wouldn't be better to
 > introduce a new function to pass to the caller active_banks and
 > nr_active_banks?

Revisited. I think it is fine how it is for now and we reconsider
later. Only thing I want to remark is that use should use kcalloc()
instead of kalloc_array() + memset().
---

See https://lkml.org/lkml/2018/11/13/1059.

How we should proceed?

Thanks

Roberto


> /Jarkko
> 

-- 
HUAWEI TECHNOLOGIES Duesseldorf GmbH, HRB 56063
Managing Director: Bo PENG, Jian LI, Yanli SHI