[PATCH v5 00/11] Regression fixes/clean ups in the Qualcomm crypto engine driver

[PATCH v5 00/11] Regression fixes/clean ups in the Qualcomm crypto engine driver

[Thara Gopinath]

This patch series is a result of running kernel crypto fuzz tests (by
enabling CONFIG_CRYPTO_MANAGER_EXTRA_TESTS) on the transformations
currently supported via the Qualcomm crypto engine on sdm845.  The first
nine patches are fixes for various regressions found during testing. The
last two patches are minor clean ups of unused variable and parameters.

v4->v5:
	- Fixed build warning/error in patch for wrong assignment of const
	  pointer as reported by kernel test robot <lkp@intel.com>.
	- Rebased to 5.11-rc6.
v3->v4:
	- Fixed the bug where only two bytes of byte_count were getting
	  saved and restored instead of all eight bytes. Thanks Bjorn for
	  catching this.
	- Split patch 3 "Fix regressions found during fuzz testing" into
	  6 patches as requested by Bjorn.
	- Dropped crypto from all subject headers.
	- Rebased to 5.11-rc5
v2->v3:
        - Made the comparison between keys to check if any two keys are
          same for triple des algorithms constant-time as per
          Nym Seddon's suggestion.
        - Rebased to 5.11-rc4.
v1->v2:
        - Introduced custom struct qce_sha_saved_state to store and restore
          partial sha transformation.
        - Rebased to 5.11-rc3.

Thara Gopinath (11):
  crypto: qce: sha: Restore/save ahash state with custom struct in
    export/import
  crypto: qce: sha: Hold back a block of data to be transferred as part
    of final
  crypto: qce: skcipher: Return unsupported if key1 and key 2 are same
    for AES XTS algorithm
  crypto: qce: skcipher: Return unsupported if any three keys are same
    for DES3 algorithms
  crypto: qce: skcipher: Return error for zero length messages
  crypto: qce: skcipher: Return error for non-blocksize data(ECB/CBC
    algorithms)
  crypto: qce: skcipher: Set ivsize to 0 for ecb(aes)
  crypto: qce: skcipher: Improve the conditions for requesting AES
    fallback cipher
  crypto: qce: common: Set data unit size to message length for AES XTS
    transformation
  crypto: qce: Remover src_tbl from qce_cipher_reqctx
  crypto: qce: Remove totallen and offset in qce_start

 drivers/crypto/qce/cipher.h   |   1 -
 drivers/crypto/qce/common.c   |  25 +++---
 drivers/crypto/qce/common.h   |   3 +-
 drivers/crypto/qce/sha.c      | 143 +++++++++++++---------------------
 drivers/crypto/qce/skcipher.c |  72 ++++++++++++++---
 5 files changed, 129 insertions(+), 115 deletions(-)

-- 
2.25.1

[PATCH v5 01/11] crypto: qce: sha: Restore/save ahash state with custom struct in export/import

[Thara Gopinath]

Export and import interfaces save and restore partial transformation
states. The partial states were being stored and restored in struct
sha1_state for sha1/hmac(sha1) transformations and sha256_state for
sha256/hmac(sha256) transformations.This led to a bunch of corner cases
where improper state was being stored and restored. A few of the corner
cases that turned up during testing are:

- wrong byte_count restored if export/import is called twice without h/w
transaction in between
- wrong buflen restored back if the pending buffer
length is exactly the block size.
- wrong state restored if buffer length is 0.

To fix these issues, save and restore the partial transformation state
using the newly introduced qce_sha_saved_state struct. This ensures that
all the pieces required to properly restart the transformation is captured
and restored back

Signed-off-by: Thara Gopinath <thara.gopinath@linaro.org>
---

v4->v5:
	- Fixed build warning/error in patch for wrong assignment of const
	  pointer as reported by kernel test robot <lkp@intel.com>.
v1->v2:
	- Introduced custom struct qce_sha_saved_state to store and
	  restore partial sha transformation. v1 was re-using
	  qce_sha_reqctx to save and restore partial states and this
	  could lead to potential memcpy issues around pointer copying.


 drivers/crypto/qce/sha.c | 122 +++++++++++----------------------------
 1 file changed, 34 insertions(+), 88 deletions(-)

diff --git a/drivers/crypto/qce/sha.c b/drivers/crypto/qce/sha.c
index 61c418c12345..7da562dca740 100644
--- a/drivers/crypto/qce/sha.c
+++ b/drivers/crypto/qce/sha.c
@@ -12,9 +12,15 @@
 #include "core.h"
 #include "sha.h"
 
-/* crypto hw padding constant for first operation */
-#define SHA_PADDING		64
-#define SHA_PADDING_MASK	(SHA_PADDING - 1)
+struct qce_sha_saved_state {
+	u8 pending_buf[QCE_SHA_MAX_BLOCKSIZE];
+	u8 partial_digest[QCE_SHA_MAX_DIGESTSIZE];
+	__be32 byte_count[2];
+	unsigned int pending_buflen;
+	unsigned int flags;
+	u64 count;
+	bool first_blk;
+};
 
 static LIST_HEAD(ahash_algs);
 
@@ -139,97 +145,37 @@ static int qce_ahash_init(struct ahash_request *req)
 
 static int qce_ahash_export(struct ahash_request *req, void *out)
 {
-	struct crypto_ahash *ahash = crypto_ahash_reqtfm(req);
 	struct qce_sha_reqctx *rctx = ahash_request_ctx(req);
-	unsigned long flags = rctx->flags;
-	unsigned int digestsize = crypto_ahash_digestsize(ahash);
-	unsigned int blocksize =
-			crypto_tfm_alg_blocksize(crypto_ahash_tfm(ahash));
-
-	if (IS_SHA1(flags) || IS_SHA1_HMAC(flags)) {
-		struct sha1_state *out_state = out;
-
-		out_state->count = rctx->count;
-		qce_cpu_to_be32p_array((__be32 *)out_state->state,
-				       rctx->digest, digestsize);
-		memcpy(out_state->buffer, rctx->buf, blocksize);
-	} else if (IS_SHA256(flags) || IS_SHA256_HMAC(flags)) {
-		struct sha256_state *out_state = out;
-
-		out_state->count = rctx->count;
-		qce_cpu_to_be32p_array((__be32 *)out_state->state,
-				       rctx->digest, digestsize);
-		memcpy(out_state->buf, rctx->buf, blocksize);
-	} else {
-		return -EINVAL;
-	}
-
-	return 0;
-}
-
-static int qce_import_common(struct ahash_request *req, u64 in_count,
-			     const u32 *state, const u8 *buffer, bool hmac)
-{
-	struct crypto_ahash *ahash = crypto_ahash_reqtfm(req);
-	struct qce_sha_reqctx *rctx = ahash_request_ctx(req);
-	unsigned int digestsize = crypto_ahash_digestsize(ahash);
-	unsigned int blocksize;
-	u64 count = in_count;
-
-	blocksize = crypto_tfm_alg_blocksize(crypto_ahash_tfm(ahash));
-	rctx->count = in_count;
-	memcpy(rctx->buf, buffer, blocksize);
-
-	if (in_count <= blocksize) {
-		rctx->first_blk = 1;
-	} else {
-		rctx->first_blk = 0;
-		/*
-		 * For HMAC, there is a hardware padding done when first block
-		 * is set. Therefore the byte_count must be incremened by 64
-		 * after the first block operation.
-		 */
-		if (hmac)
-			count += SHA_PADDING;
-	}
+	struct qce_sha_saved_state *export_state = out;
 
-	rctx->byte_count[0] = (__force __be32)(count & ~SHA_PADDING_MASK);
-	rctx->byte_count[1] = (__force __be32)(count >> 32);
-	qce_cpu_to_be32p_array((__be32 *)rctx->digest, (const u8 *)state,
-			       digestsize);
-	rctx->buflen = (unsigned int)(in_count & (blocksize - 1));
+	memcpy(export_state->pending_buf, rctx->buf, rctx->buflen);
+	memcpy(export_state->partial_digest, rctx->digest, sizeof(rctx->digest));
+	export_state->byte_count[0] = rctx->byte_count[0];
+	export_state->byte_count[1] = rctx->byte_count[1];
+	export_state->pending_buflen = rctx->buflen;
+	export_state->count = rctx->count;
+	export_state->first_blk = rctx->first_blk;
+	export_state->flags = rctx->flags;
 
 	return 0;
 }
 
 static int qce_ahash_import(struct ahash_request *req, const void *in)
 {
-	struct qce_sha_reqctx *rctx;
-	unsigned long flags;
-	bool hmac;
-	int ret;
-
-	ret = qce_ahash_init(req);
-	if (ret)
-		return ret;
-
-	rctx = ahash_request_ctx(req);
-	flags = rctx->flags;
-	hmac = IS_SHA_HMAC(flags);
-
-	if (IS_SHA1(flags) || IS_SHA1_HMAC(flags)) {
-		const struct sha1_state *state = in;
-
-		ret = qce_import_common(req, state->count, state->state,
-					state->buffer, hmac);
-	} else if (IS_SHA256(flags) || IS_SHA256_HMAC(flags)) {
-		const struct sha256_state *state = in;
+	struct qce_sha_reqctx *rctx = ahash_request_ctx(req);
+	const struct qce_sha_saved_state *import_state = in;
 
-		ret = qce_import_common(req, state->count, state->state,
-					state->buf, hmac);
-	}
+	memset(rctx, 0, sizeof(*rctx));
+	rctx->count = import_state->count;
+	rctx->buflen = import_state->pending_buflen;
+	rctx->first_blk = import_state->first_blk;
+	rctx->flags = import_state->flags;
+	rctx->byte_count[0] = import_state->byte_count[0];
+	rctx->byte_count[1] = import_state->byte_count[1];
+	memcpy(rctx->buf, import_state->pending_buf, rctx->buflen);
+	memcpy(rctx->digest, import_state->partial_digest, sizeof(rctx->digest));
 
-	return ret;
+	return 0;
 }
 
 static int qce_ahash_update(struct ahash_request *req)
@@ -450,7 +396,7 @@ static const struct qce_ahash_def ahash_def[] = {
 		.drv_name	= "sha1-qce",
 		.digestsize	= SHA1_DIGEST_SIZE,
 		.blocksize	= SHA1_BLOCK_SIZE,
-		.statesize	= sizeof(struct sha1_state),
+		.statesize	= sizeof(struct qce_sha_saved_state),
 		.std_iv		= std_iv_sha1,
 	},
 	{
@@ -459,7 +405,7 @@ static const struct qce_ahash_def ahash_def[] = {
 		.drv_name	= "sha256-qce",
 		.digestsize	= SHA256_DIGEST_SIZE,
 		.blocksize	= SHA256_BLOCK_SIZE,
-		.statesize	= sizeof(struct sha256_state),
+		.statesize	= sizeof(struct qce_sha_saved_state),
 		.std_iv		= std_iv_sha256,
 	},
 	{
@@ -468,7 +414,7 @@ static const struct qce_ahash_def ahash_def[] = {
 		.drv_name	= "hmac-sha1-qce",
 		.digestsize	= SHA1_DIGEST_SIZE,
 		.blocksize	= SHA1_BLOCK_SIZE,
-		.statesize	= sizeof(struct sha1_state),
+		.statesize	= sizeof(struct qce_sha_saved_state),
 		.std_iv		= std_iv_sha1,
 	},
 	{
@@ -477,7 +423,7 @@ static const struct qce_ahash_def ahash_def[] = {
 		.drv_name	= "hmac-sha256-qce",
 		.digestsize	= SHA256_DIGEST_SIZE,
 		.blocksize	= SHA256_BLOCK_SIZE,
-		.statesize	= sizeof(struct sha256_state),
+		.statesize	= sizeof(struct qce_sha_saved_state),
 		.std_iv		= std_iv_sha256,
 	},
 };
-- 
2.25.1

[PATCH v5 02/11] crypto: qce: sha: Hold back a block of data to be transferred as part of final

[Thara Gopinath]

If the available data to transfer is exactly a multiple of block size, save
the last block to be transferred in qce_ahash_final (with the last block
bit set) if this is indeed the end of data stream. If not this saved block
will be transferred as part of next update. If this block is not held back
and if this is indeed the end of data stream, the digest obtained will be
wrong since qce_ahash_final will see that rctx->buflen is 0 and return
doing nothing which in turn means that a digest will not be copied to the
destination result buffer.  qce_ahash_final cannot be made to alter this
behavior and allowed to proceed if rctx->buflen is 0 because the crypto
engine BAM does not allow for zero length transfers.

Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Thara Gopinath <thara.gopinath@linaro.org>
---
 drivers/crypto/qce/sha.c | 19 +++++++++++++++++++
 1 file changed, 19 insertions(+)

diff --git a/drivers/crypto/qce/sha.c b/drivers/crypto/qce/sha.c
index 7da562dca740..2813c9a27a6e 100644
--- a/drivers/crypto/qce/sha.c
+++ b/drivers/crypto/qce/sha.c
@@ -216,6 +216,25 @@ static int qce_ahash_update(struct ahash_request *req)
 
 	/* calculate how many bytes will be hashed later */
 	hash_later = total % blocksize;
+
+	/*
+	 * At this point, there is more than one block size of data.  If
+	 * the available data to transfer is exactly a multiple of block
+	 * size, save the last block to be transferred in qce_ahash_final
+	 * (with the last block bit set) if this is indeed the end of data
+	 * stream. If not this saved block will be transferred as part of
+	 * next update. If this block is not held back and if this is
+	 * indeed the end of data stream, the digest obtained will be wrong
+	 * since qce_ahash_final will see that rctx->buflen is 0 and return
+	 * doing nothing which in turn means that a digest will not be
+	 * copied to the destination result buffer.  qce_ahash_final cannot
+	 * be made to alter this behavior and allowed to proceed if
+	 * rctx->buflen is 0 because the crypto engine BAM does not allow
+	 * for zero length transfers.
+	 */
+	if (!hash_later)
+		hash_later = blocksize;
+
 	if (hash_later) {
 		unsigned int src_offset = req->nbytes - hash_later;
 		scatterwalk_map_and_copy(rctx->buf, req->src, src_offset,
-- 
2.25.1

[PATCH v5 03/11] crypto: qce: skcipher: Return unsupported if key1 and key 2 are same for AES XTS algorithm

[Thara Gopinath]

Crypto engine does not support key1 = key2 for AES XTS algorithm; the
operation hangs the engines.  Return -EINVAL in case key1 and key2 are the
same.

Signed-off-by: Thara Gopinath <thara.gopinath@linaro.org>
---
 drivers/crypto/qce/skcipher.c | 19 ++++++++++++++++++-
 1 file changed, 18 insertions(+), 1 deletion(-)

diff --git a/drivers/crypto/qce/skcipher.c b/drivers/crypto/qce/skcipher.c
index a2d3da0ad95f..12955dcd53dd 100644
--- a/drivers/crypto/qce/skcipher.c
+++ b/drivers/crypto/qce/skcipher.c
@@ -167,16 +167,33 @@ static int qce_skcipher_setkey(struct crypto_skcipher *ablk, const u8 *key,
 	struct crypto_tfm *tfm = crypto_skcipher_tfm(ablk);
 	struct qce_cipher_ctx *ctx = crypto_tfm_ctx(tfm);
 	unsigned long flags = to_cipher_tmpl(ablk)->alg_flags;
+	unsigned int __keylen;
 	int ret;
 
 	if (!key || !keylen)
 		return -EINVAL;
 
-	switch (IS_XTS(flags) ? keylen >> 1 : keylen) {
+	/*
+	 * AES XTS key1 = key2 not supported by crypto engine.
+	 * Revisit to request a fallback cipher in this case.
+	 */
+	if (IS_XTS(flags)) {
+		__keylen = keylen >> 1;
+		if (!memcmp(key, key + __keylen, __keylen))
+			return -ENOKEY;
+	} else {
+		__keylen = keylen;
+	}
+
+	switch (__keylen) {
 	case AES_KEYSIZE_128:
 	case AES_KEYSIZE_256:
 		memcpy(ctx->enc_key, key, keylen);
 		break;
+	case AES_KEYSIZE_192:
+		break;
+	default:
+		return -EINVAL;
 	}
 
 	ret = crypto_skcipher_setkey(ctx->fallback, key, keylen);
-- 
2.25.1

[PATCH v5 04/11] crypto: qce: skcipher: Return unsupported if any three keys are same for DES3 algorithms

[Thara Gopinath]

Return unsupported if any three keys are same for DES3 algorithms
since CE does not support this and the operation causes the engine to
hang.

Signed-off-by: Thara Gopinath <thara.gopinath@linaro.org>
---
 drivers/crypto/qce/skcipher.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/drivers/crypto/qce/skcipher.c b/drivers/crypto/qce/skcipher.c
index 12955dcd53dd..de1f37ed4ee6 100644
--- a/drivers/crypto/qce/skcipher.c
+++ b/drivers/crypto/qce/skcipher.c
@@ -221,12 +221,27 @@ static int qce_des3_setkey(struct crypto_skcipher *ablk, const u8 *key,
 			   unsigned int keylen)
 {
 	struct qce_cipher_ctx *ctx = crypto_skcipher_ctx(ablk);
+	u32 _key[6];
 	int err;
 
 	err = verify_skcipher_des3_key(ablk, key);
 	if (err)
 		return err;
 
+	/*
+	 * The crypto engine does not support any two keys
+	 * being the same for triple des algorithms. The
+	 * verify_skcipher_des3_key does not check for all the
+	 * below conditions. Return -ENOKEY in case any two keys
+	 * are the same. Revisit to see if a fallback cipher
+	 * is needed to handle this condition.
+	 */
+	memcpy(_key, key, DES3_EDE_KEY_SIZE);
+	if (!((_key[0] ^ _key[2]) | (_key[1] ^ _key[3])) |
+	    !((_key[2] ^ _key[4]) | (_key[3] ^ _key[5])) |
+	    !((_key[0] ^ _key[4]) | (_key[1] ^ _key[5])))
+		return -ENOKEY;
+
 	ctx->enc_keylen = keylen;
 	memcpy(ctx->enc_key, key, keylen);
 	return 0;
-- 
2.25.1

[PATCH v5 05/11] crypto: qce: skcipher: Return error for zero length messages

[Thara Gopinath]

Crypto engine BAM dma does not support 0 length data. Return unsupported
if zero length messages are passed for transformation.

Signed-off-by: Thara Gopinath <thara.gopinath@linaro.org>
---
 drivers/crypto/qce/skcipher.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/crypto/qce/skcipher.c b/drivers/crypto/qce/skcipher.c
index de1f37ed4ee6..331b3c3a5b59 100644
--- a/drivers/crypto/qce/skcipher.c
+++ b/drivers/crypto/qce/skcipher.c
@@ -8,6 +8,7 @@
 #include <linux/interrupt.h>
 #include <linux/moduleparam.h>
 #include <linux/types.h>
+#include <linux/errno.h>
 #include <crypto/aes.h>
 #include <crypto/internal/des.h>
 #include <crypto/internal/skcipher.h>
@@ -260,6 +261,10 @@ static int qce_skcipher_crypt(struct skcipher_request *req, int encrypt)
 	rctx->flags |= encrypt ? QCE_ENCRYPT : QCE_DECRYPT;
 	keylen = IS_XTS(rctx->flags) ? ctx->enc_keylen >> 1 : ctx->enc_keylen;
 
+	/* CE does not handle 0 length messages */
+	if (!req->cryptlen)
+		return -EOPNOTSUPP;
+
 	/* qce is hanging when AES-XTS request len > QCE_SECTOR_SIZE and
 	 * is not a multiple of it; pass such requests to the fallback
 	 */
-- 
2.25.1

Re: [PATCH v5 05/11] crypto: qce: skcipher: Return error for zero length messages

[Eric Biggers]

On Thu, Feb 04, 2021 at 04:43:53PM -0500, Thara Gopinath wrote:
> Crypto engine BAM dma does not support 0 length data. Return unsupported
> if zero length messages are passed for transformation.
> 
> Signed-off-by: Thara Gopinath <thara.gopinath@linaro.org>
> ---
>  drivers/crypto/qce/skcipher.c | 5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/drivers/crypto/qce/skcipher.c b/drivers/crypto/qce/skcipher.c
> index de1f37ed4ee6..331b3c3a5b59 100644
> --- a/drivers/crypto/qce/skcipher.c
> +++ b/drivers/crypto/qce/skcipher.c
> @@ -8,6 +8,7 @@
>  #include <linux/interrupt.h>
>  #include <linux/moduleparam.h>
>  #include <linux/types.h>
> +#include <linux/errno.h>
>  #include <crypto/aes.h>
>  #include <crypto/internal/des.h>
>  #include <crypto/internal/skcipher.h>
> @@ -260,6 +261,10 @@ static int qce_skcipher_crypt(struct skcipher_request *req, int encrypt)
>  	rctx->flags |= encrypt ? QCE_ENCRYPT : QCE_DECRYPT;
>  	keylen = IS_XTS(rctx->flags) ? ctx->enc_keylen >> 1 : ctx->enc_keylen;
>  
> +	/* CE does not handle 0 length messages */
> +	if (!req->cryptlen)
> +		return -EOPNOTSUPP;
> +

For the algorithms in question, the correct behavior is to return 0.

Aren't the tests catching that difference?

- Eric

Re: [PATCH v5 05/11] crypto: qce: skcipher: Return error for zero length messages

[Thara Gopinath]

Hi Eric,

On 2/4/21 5:48 PM, Eric Biggers wrote:
> On Thu, Feb 04, 2021 at 04:43:53PM -0500, Thara Gopinath wrote:
>> Crypto engine BAM dma does not support 0 length data. Return unsupported
>> if zero length messages are passed for transformation.
>>
>> Signed-off-by: Thara Gopinath <thara.gopinath@linaro.org>
>> ---
>>   drivers/crypto/qce/skcipher.c | 5 +++++
>>   1 file changed, 5 insertions(+)
>>
>> diff --git a/drivers/crypto/qce/skcipher.c b/drivers/crypto/qce/skcipher.c
>> index de1f37ed4ee6..331b3c3a5b59 100644
>> --- a/drivers/crypto/qce/skcipher.c
>> +++ b/drivers/crypto/qce/skcipher.c
>> @@ -8,6 +8,7 @@
>>   #include <linux/interrupt.h>
>>   #include <linux/moduleparam.h>
>>   #include <linux/types.h>
>> +#include <linux/errno.h>
>>   #include <crypto/aes.h>
>>   #include <crypto/internal/des.h>
>>   #include <crypto/internal/skcipher.h>
>> @@ -260,6 +261,10 @@ static int qce_skcipher_crypt(struct skcipher_request *req, int encrypt)
>>   	rctx->flags |= encrypt ? QCE_ENCRYPT : QCE_DECRYPT;
>>   	keylen = IS_XTS(rctx->flags) ? ctx->enc_keylen >> 1 : ctx->enc_keylen;
>>   
>> +	/* CE does not handle 0 length messages */
>> +	if (!req->cryptlen)
>> +		return -EOPNOTSUPP;
>> +
> 
> For the algorithms in question, the correct behavior is to return 0.

What do you mean? The driver should return a 0 ?

> 
> Aren't the tests catching that difference?

I was anyways planning on sending an email to the list with these 
queries. But since you asked,  these are my observations with fuzz 
testing which I have been doing quite a bit now (I am also working on 
adding a few qualcomm AEAD algorithms support in mainline).

- if the generic algorithm supports 0 length messages and the 
transformation I am testing does not, the test framework throws an error 
and stops.
- key support mismatch between the generic algorithm vs my algorithm 
/engine also does the same thing.For eg, Qualcomm CE engine does not 
support any three keys being same for triple des algorithms. Where as a 
two key 3des is a valid scenario for generic algorithm(k1=k3). Another 
example is hardware engine not supporting AES192.

How are these scenarios usually handled ? Why not allow the test 
framework to proceed with the testing if the algorithm does not support 
a particular scenario ?

> 
> - Eric
> 
-- 
Warm Regards
Thara

Re: [PATCH v5 05/11] crypto: qce: skcipher: Return error for zero length messages

[Eric Biggers]

On Thu, Feb 04, 2021 at 07:09:53PM -0500, Thara Gopinath wrote:
> > > @@ -260,6 +261,10 @@ static int qce_skcipher_crypt(struct skcipher_request *req, int encrypt)
> > >   	rctx->flags |= encrypt ? QCE_ENCRYPT : QCE_DECRYPT;
> > >   	keylen = IS_XTS(rctx->flags) ? ctx->enc_keylen >> 1 : ctx->enc_keylen;
> > > +	/* CE does not handle 0 length messages */
> > > +	if (!req->cryptlen)
> > > +		return -EOPNOTSUPP;
> > > +
> > 
> > For the algorithms in question, the correct behavior is to return 0.
> 
> What do you mean? The driver should return a 0 ?

Yes, there is nothing to do for empty inputs, so just return 0 (success).

> > Aren't the tests catching that difference?
> 
> I was anyways planning on sending an email to the list with these queries.
> But since you asked,  these are my observations with fuzz testing which I
> have been doing quite a bit now (I am also working on adding a few qualcomm
> AEAD algorithms support in mainline).
> 
> - if the generic algorithm supports 0 length messages and the transformation
> I am testing does not, the test framework throws an error and stops.
> - key support mismatch between the generic algorithm vs my algorithm /engine
> also does the same thing.For eg, Qualcomm CE engine does not support any
> three keys being same for triple des algorithms. Where as a two key 3des is
> a valid scenario for generic algorithm(k1=k3). Another example is hardware
> engine not supporting AES192.
> 
> How are these scenarios usually handled ? Why not allow the test framework
> to proceed with the testing if the algorithm does not support a particular
> scenario ?

Omitting support for certain inputs isn't allowed.  Anyone in the kernel who
wants to use a particular algorithm could get this driver for it, and if they
happen to use inputs which the driver decided not to support, things will break.

The way that drivers handle this is to use a fallback cipher for inputs they
don't support.

- Eric

Re: [PATCH v5 05/11] crypto: qce: skcipher: Return error for zero length messages

[Thara Gopinath]

On 2/4/21 7:26 PM, Eric Biggers wrote:
> On Thu, Feb 04, 2021 at 07:09:53PM -0500, Thara Gopinath wrote:
>>>> @@ -260,6 +261,10 @@ static int qce_skcipher_crypt(struct skcipher_request *req, int encrypt)
>>>>    	rctx->flags |= encrypt ? QCE_ENCRYPT : QCE_DECRYPT;
>>>>    	keylen = IS_XTS(rctx->flags) ? ctx->enc_keylen >> 1 : ctx->enc_keylen;
>>>> +	/* CE does not handle 0 length messages */
>>>> +	if (!req->cryptlen)
>>>> +		return -EOPNOTSUPP;
>>>> +
>>>
>>> For the algorithms in question, the correct behavior is to return 0.
>>
>> What do you mean? The driver should return a 0 ?

Ok. I will re-spin the series once more with this change..

> 
> Yes, there is nothing to do for empty inputs, so just return 0 (success).
> 
>>> Aren't the tests catching that difference?
>>
>> I was anyways planning on sending an email to the list with these queries.
>> But since you asked,  these are my observations with fuzz testing which I
>> have been doing quite a bit now (I am also working on adding a few qualcomm
>> AEAD algorithms support in mainline).
>>
>> - if the generic algorithm supports 0 length messages and the transformation
>> I am testing does not, the test framework throws an error and stops.
>> - key support mismatch between the generic algorithm vs my algorithm /engine
>> also does the same thing.For eg, Qualcomm CE engine does not support any
>> three keys being same for triple des algorithms. Where as a two key 3des is
>> a valid scenario for generic algorithm(k1=k3). Another example is hardware
>> engine not supporting AES192.
>>
>> How are these scenarios usually handled ? Why not allow the test framework
>> to proceed with the testing if the algorithm does not support a particular
>> scenario ?
> 
> Omitting support for certain inputs isn't allowed.  Anyone in the kernel who
> wants to use a particular algorithm could get this driver for it, and if they
> happen to use inputs which the driver decided not to support, things will break.

Ya sounds reasonable.

> 
> The way that drivers handle this is to use a fallback cipher for inputs they
> don't support.

Ok. So I will add this to my todo and make sure to have fallback ciphers 
for all the non-supported inputs. I will send this as a separate series 
and not this one.

In this case, though not supporting 0 length messages for encryption is 
valid. I don't think I have to have a fallback for this. I could have 
sworn that the test framework throws up an error for this. But I have 
been testing a lot and may be I am just confused. I will double check this.


> 
> - Eric
> 

-- 
Warm Regards
Thara

[PATCH v5 06/11] crypto: qce: skcipher: Return error for non-blocksize data(ECB/CBC algorithms)

[Thara Gopinath]

ECB/CBC encryption/decryption requires the data to be blocksize aligned.
Crypto engine hangs on non-block sized operations for these algorithms.
Return invalid data if data size is not blocksize aligned for these
algorithms.

Signed-off-by: Thara Gopinath <thara.gopinath@linaro.org>
---
 drivers/crypto/qce/skcipher.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/drivers/crypto/qce/skcipher.c b/drivers/crypto/qce/skcipher.c
index 331b3c3a5b59..28bea9584c33 100644
--- a/drivers/crypto/qce/skcipher.c
+++ b/drivers/crypto/qce/skcipher.c
@@ -254,6 +254,7 @@ static int qce_skcipher_crypt(struct skcipher_request *req, int encrypt)
 	struct qce_cipher_ctx *ctx = crypto_skcipher_ctx(tfm);
 	struct qce_cipher_reqctx *rctx = skcipher_request_ctx(req);
 	struct qce_alg_template *tmpl = to_cipher_tmpl(tfm);
+	unsigned int blocksize = crypto_skcipher_blocksize(tfm);
 	int keylen;
 	int ret;
 
@@ -265,6 +266,17 @@ static int qce_skcipher_crypt(struct skcipher_request *req, int encrypt)
 	if (!req->cryptlen)
 		return -EOPNOTSUPP;
 
+	/*
+	 * ECB and CBC algorithms require message lengths to be
+	 * multiples of block size.
+	 * TODO: The spec says AES CBC mode for certain versions
+	 * of crypto engine can handle partial blocks as well.
+	 * Test and enable such messages.
+	 */
+	if (IS_ECB(rctx->flags) || IS_CBC(rctx->flags))
+		if (!IS_ALIGNED(req->cryptlen, blocksize))
+			return -EINVAL;
+
 	/* qce is hanging when AES-XTS request len > QCE_SECTOR_SIZE and
 	 * is not a multiple of it; pass such requests to the fallback
 	 */
-- 
2.25.1

Re: [PATCH v5 06/11] crypto: qce: skcipher: Return error for non-blocksize data(ECB/CBC algorithms)

[Eric Biggers]

On Thu, Feb 04, 2021 at 04:43:54PM -0500, Thara Gopinath wrote:
> +	/*
> +	 * ECB and CBC algorithms require message lengths to be
> +	 * multiples of block size.
> +	 * TODO: The spec says AES CBC mode for certain versions
> +	 * of crypto engine can handle partial blocks as well.
> +	 * Test and enable such messages.
> +	 */
> +	if (IS_ECB(rctx->flags) || IS_CBC(rctx->flags))
> +		if (!IS_ALIGNED(req->cryptlen, blocksize))
> +			return -EINVAL;

CBC by definition only operates on full blocks, so the TODO doesn't make sense.
Is the partial block support really CTS-CBC?

- Eric

Re: [PATCH v5 06/11] crypto: qce: skcipher: Return error for non-blocksize data(ECB/CBC algorithms)

[Thara Gopinath]

On 2/4/21 5:50 PM, Eric Biggers wrote:
> On Thu, Feb 04, 2021 at 04:43:54PM -0500, Thara Gopinath wrote:
>> +	/*
>> +	 * ECB and CBC algorithms require message lengths to be
>> +	 * multiples of block size.
>> +	 * TODO: The spec says AES CBC mode for certain versions
>> +	 * of crypto engine can handle partial blocks as well.
>> +	 * Test and enable such messages.
>> +	 */
>> +	if (IS_ECB(rctx->flags) || IS_CBC(rctx->flags))
>> +		if (!IS_ALIGNED(req->cryptlen, blocksize))
>> +			return -EINVAL;
> 
> CBC by definition only operates on full blocks, so the TODO doesn't make sense.
> Is the partial block support really CTS-CBC?

Ya you are right. It should be CTS-CBC and not AES CBC. Though the spec 
is quite fuzzy about this part.

I can remove the comment and spin the next version or just leave it 
there for now and remove it later.

> 
> - Eric
> 

-- 
Warm Regards
Thara

[PATCH v5 07/11] crypto: qce: skcipher: Set ivsize to 0 for ecb(aes)

[Thara Gopinath]

ECB transformations do not have an IV and hence set the ivsize to 0 for
ecb(aes).

Signed-off-by: Thara Gopinath <thara.gopinath@linaro.org>
---
 drivers/crypto/qce/skcipher.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/crypto/qce/skcipher.c b/drivers/crypto/qce/skcipher.c
index 28bea9584c33..10e85b1fc0fd 100644
--- a/drivers/crypto/qce/skcipher.c
+++ b/drivers/crypto/qce/skcipher.c
@@ -356,7 +356,7 @@ static const struct qce_skcipher_def skcipher_def[] = {
 		.name		= "ecb(aes)",
 		.drv_name	= "ecb-aes-qce",
 		.blocksize	= AES_BLOCK_SIZE,
-		.ivsize		= AES_BLOCK_SIZE,
+		.ivsize		= 0,
 		.min_keysize	= AES_MIN_KEY_SIZE,
 		.max_keysize	= AES_MAX_KEY_SIZE,
 	},
-- 
2.25.1

[PATCH v5 08/11] crypto: qce: skcipher: Improve the conditions for requesting AES fallback cipher

[Thara Gopinath]

The following are the conditions for requesting AES fallback cipher.
	-  AES-192
	- AES-XTS request with len <= 512 byte (Allow messages of length
	  less than 512 bytes for all other AES encryption algorithms other
	  than AES XTS)
	- AES-XTS request with len > QCE_SECTOR_SIZE and is not a multiple
	  of it

Signed-off-by: Thara Gopinath <thara.gopinath@linaro.org>
---
 drivers/crypto/qce/skcipher.c | 17 +++++++++++------
 1 file changed, 11 insertions(+), 6 deletions(-)

diff --git a/drivers/crypto/qce/skcipher.c b/drivers/crypto/qce/skcipher.c
index 10e85b1fc0fd..8599250946b7 100644
--- a/drivers/crypto/qce/skcipher.c
+++ b/drivers/crypto/qce/skcipher.c
@@ -277,14 +277,19 @@ static int qce_skcipher_crypt(struct skcipher_request *req, int encrypt)
 		if (!IS_ALIGNED(req->cryptlen, blocksize))
 			return -EINVAL;
 
-	/* qce is hanging when AES-XTS request len > QCE_SECTOR_SIZE and
-	 * is not a multiple of it; pass such requests to the fallback
+	/*
+	 * Conditions for requesting a fallback cipher
+	 * AES-192 (not supported by crypto engine (CE))
+	 * AES-XTS request with len <= 512 byte (not recommended to use CE)
+	 * AES-XTS request with len > QCE_SECTOR_SIZE and
+	 * is not a multiple of it.(Revisit this condition to check if it is
+	 * needed in all versions of CE)
 	 */
 	if (IS_AES(rctx->flags) &&
-	    (((keylen != AES_KEYSIZE_128 && keylen != AES_KEYSIZE_256) ||
-	      req->cryptlen <= aes_sw_max_len) ||
-	     (IS_XTS(rctx->flags) && req->cryptlen > QCE_SECTOR_SIZE &&
-	      req->cryptlen % QCE_SECTOR_SIZE))) {
+	    ((keylen != AES_KEYSIZE_128 && keylen != AES_KEYSIZE_256) ||
+	    (IS_XTS(rctx->flags) && ((req->cryptlen <= aes_sw_max_len) ||
+	    (req->cryptlen > QCE_SECTOR_SIZE &&
+	    req->cryptlen % QCE_SECTOR_SIZE))))) {
 		skcipher_request_set_tfm(&rctx->fallback_req, ctx->fallback);
 		skcipher_request_set_callback(&rctx->fallback_req,
 					      req->base.flags,
-- 
2.25.1

[PATCH v5 09/11] crypto: qce: common: Set data unit size to message length for AES XTS transformation

[Thara Gopinath]

Set the register REG_ENCR_XTS_DU_SIZE to cryptlen for AES XTS
transformation. Anything else causes the engine to return back
wrong results.

Acked-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Thara Gopinath <thara.gopinath@linaro.org>
---
 drivers/crypto/qce/common.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/crypto/qce/common.c b/drivers/crypto/qce/common.c
index a73db2a5637f..f7bc701a4aa2 100644
--- a/drivers/crypto/qce/common.c
+++ b/drivers/crypto/qce/common.c
@@ -295,15 +295,15 @@ static void qce_xtskey(struct qce_device *qce, const u8 *enckey,
 {
 	u32 xtskey[QCE_MAX_CIPHER_KEY_SIZE / sizeof(u32)] = {0};
 	unsigned int xtsklen = enckeylen / (2 * sizeof(u32));
-	unsigned int xtsdusize;
 
 	qce_cpu_to_be32p_array((__be32 *)xtskey, enckey + enckeylen / 2,
 			       enckeylen / 2);
 	qce_write_array(qce, REG_ENCR_XTS_KEY0, xtskey, xtsklen);
 
-	/* xts du size 512B */
-	xtsdusize = min_t(u32, QCE_SECTOR_SIZE, cryptlen);
-	qce_write(qce, REG_ENCR_XTS_DU_SIZE, xtsdusize);
+	/* Set data unit size to cryptlen. Anything else causes
+	 * crypto engine to return back incorrect results.
+	 */
+	qce_write(qce, REG_ENCR_XTS_DU_SIZE, cryptlen);
 }
 
 static int qce_setup_regs_skcipher(struct crypto_async_request *async_req,
-- 
2.25.1

[PATCH v5 10/11] crypto: qce: Remover src_tbl from qce_cipher_reqctx

[Thara Gopinath]

src_table is unused and hence remove it from struct qce_cipher_reqctx

Signed-off-by: Thara Gopinath <thara.gopinath@linaro.org>
---
 drivers/crypto/qce/cipher.h | 1 -
 1 file changed, 1 deletion(-)

diff --git a/drivers/crypto/qce/cipher.h b/drivers/crypto/qce/cipher.h
index cffa9fc628ff..850f257d00f3 100644
--- a/drivers/crypto/qce/cipher.h
+++ b/drivers/crypto/qce/cipher.h
@@ -40,7 +40,6 @@ struct qce_cipher_reqctx {
 	struct scatterlist result_sg;
 	struct sg_table dst_tbl;
 	struct scatterlist *dst_sg;
-	struct sg_table src_tbl;
 	struct scatterlist *src_sg;
 	unsigned int cryptlen;
 	struct skcipher_request fallback_req;	// keep at the end
-- 
2.25.1

[PATCH v5 11/11] crypto: qce: Remove totallen and offset in qce_start

[Thara Gopinath]

totallen is used to get the size of the data to be transformed.
This is also available via nbytes or cryptlen in the qce_sha_reqctx
and qce_cipher_ctx. Similarly offset convey nothing for the supported
encryption and authentication transformations and is always 0.
Remove these two redundant parameters in qce_start.

Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Thara Gopinath <thara.gopinath@linaro.org>
---
 drivers/crypto/qce/common.c   | 17 +++++++----------
 drivers/crypto/qce/common.h   |  3 +--
 drivers/crypto/qce/sha.c      |  2 +-
 drivers/crypto/qce/skcipher.c |  2 +-
 4 files changed, 10 insertions(+), 14 deletions(-)

diff --git a/drivers/crypto/qce/common.c b/drivers/crypto/qce/common.c
index f7bc701a4aa2..dceb9579d87a 100644
--- a/drivers/crypto/qce/common.c
+++ b/drivers/crypto/qce/common.c
@@ -140,8 +140,7 @@ static u32 qce_auth_cfg(unsigned long flags, u32 key_size)
 	return cfg;
 }
 
-static int qce_setup_regs_ahash(struct crypto_async_request *async_req,
-				u32 totallen, u32 offset)
+static int qce_setup_regs_ahash(struct crypto_async_request *async_req)
 {
 	struct ahash_request *req = ahash_request_cast(async_req);
 	struct crypto_ahash *ahash = __crypto_ahash_cast(async_req->tfm);
@@ -306,8 +305,7 @@ static void qce_xtskey(struct qce_device *qce, const u8 *enckey,
 	qce_write(qce, REG_ENCR_XTS_DU_SIZE, cryptlen);
 }
 
-static int qce_setup_regs_skcipher(struct crypto_async_request *async_req,
-				     u32 totallen, u32 offset)
+static int qce_setup_regs_skcipher(struct crypto_async_request *async_req)
 {
 	struct skcipher_request *req = skcipher_request_cast(async_req);
 	struct qce_cipher_reqctx *rctx = skcipher_request_ctx(req);
@@ -367,7 +365,7 @@ static int qce_setup_regs_skcipher(struct crypto_async_request *async_req,
 
 	qce_write(qce, REG_ENCR_SEG_CFG, encr_cfg);
 	qce_write(qce, REG_ENCR_SEG_SIZE, rctx->cryptlen);
-	qce_write(qce, REG_ENCR_SEG_START, offset & 0xffff);
+	qce_write(qce, REG_ENCR_SEG_START, 0);
 
 	if (IS_CTR(flags)) {
 		qce_write(qce, REG_CNTR_MASK, ~0);
@@ -376,7 +374,7 @@ static int qce_setup_regs_skcipher(struct crypto_async_request *async_req,
 		qce_write(qce, REG_CNTR_MASK2, ~0);
 	}
 
-	qce_write(qce, REG_SEG_SIZE, totallen);
+	qce_write(qce, REG_SEG_SIZE, rctx->cryptlen);
 
 	/* get little endianness */
 	config = qce_config_reg(qce, 1);
@@ -388,17 +386,16 @@ static int qce_setup_regs_skcipher(struct crypto_async_request *async_req,
 }
 #endif
 
-int qce_start(struct crypto_async_request *async_req, u32 type, u32 totallen,
-	      u32 offset)
+int qce_start(struct crypto_async_request *async_req, u32 type)
 {
 	switch (type) {
 #ifdef CONFIG_CRYPTO_DEV_QCE_SKCIPHER
 	case CRYPTO_ALG_TYPE_SKCIPHER:
-		return qce_setup_regs_skcipher(async_req, totallen, offset);
+		return qce_setup_regs_skcipher(async_req);
 #endif
 #ifdef CONFIG_CRYPTO_DEV_QCE_SHA
 	case CRYPTO_ALG_TYPE_AHASH:
-		return qce_setup_regs_ahash(async_req, totallen, offset);
+		return qce_setup_regs_ahash(async_req);
 #endif
 	default:
 		return -EINVAL;
diff --git a/drivers/crypto/qce/common.h b/drivers/crypto/qce/common.h
index 85ba16418a04..3bc244bcca2d 100644
--- a/drivers/crypto/qce/common.h
+++ b/drivers/crypto/qce/common.h
@@ -94,7 +94,6 @@ struct qce_alg_template {
 void qce_cpu_to_be32p_array(__be32 *dst, const u8 *src, unsigned int len);
 int qce_check_status(struct qce_device *qce, u32 *status);
 void qce_get_version(struct qce_device *qce, u32 *major, u32 *minor, u32 *step);
-int qce_start(struct crypto_async_request *async_req, u32 type, u32 totallen,
-	      u32 offset);
+int qce_start(struct crypto_async_request *async_req, u32 type);
 
 #endif /* _COMMON_H_ */
diff --git a/drivers/crypto/qce/sha.c b/drivers/crypto/qce/sha.c
index 2813c9a27a6e..8e6fcf2c21cc 100644
--- a/drivers/crypto/qce/sha.c
+++ b/drivers/crypto/qce/sha.c
@@ -113,7 +113,7 @@ static int qce_ahash_async_req_handle(struct crypto_async_request *async_req)
 
 	qce_dma_issue_pending(&qce->dma);
 
-	ret = qce_start(async_req, tmpl->crypto_alg_type, 0, 0);
+	ret = qce_start(async_req, tmpl->crypto_alg_type);
 	if (ret)
 		goto error_terminate;
 
diff --git a/drivers/crypto/qce/skcipher.c b/drivers/crypto/qce/skcipher.c
index 8599250946b7..3fc0b263d498 100644
--- a/drivers/crypto/qce/skcipher.c
+++ b/drivers/crypto/qce/skcipher.c
@@ -144,7 +144,7 @@ qce_skcipher_async_req_handle(struct crypto_async_request *async_req)
 
 	qce_dma_issue_pending(&qce->dma);
 
-	ret = qce_start(async_req, tmpl->crypto_alg_type, req->cryptlen, 0);
+	ret = qce_start(async_req, tmpl->crypto_alg_type);
 	if (ret)
 		goto error_terminate;
 
-- 
2.25.1