Fix a leak on error path in qat asym

Reported-by: Salvatore Benedetto <salvatore.benedetto@intel.com>
Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

The patch centralizes the XTS key check logic into the service function
xts_check_key which is invoked from the different XTS implementations.
With this, the XTS implementations in ARM, ARM64, PPC and S390 have now
a sanity check for the XTS keys similar to the other arches.

In addition, this service function received a check to ensure that the
key != the tweak key which is mandated by FIPS 140-2 IG A.9. As the
check is not present in the standards defining XTS, it is only enforced
in FIPS mode of the kernel.

Signed-off-by: Stephan Mueller <smueller@chronox.de>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Add support for the hardware true random number generator
peripheral found on PIC32.

Signed-off-by: Joshua Henderson <joshua.henderson@microchip.com>
Signed-off-by: Purna Chandra Mandal <purna.mandal@microchip.com>
Reviewed-by: Daniel Thompson <daniel.thompson@linaro.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Document the devicetree bindings for the random number generator found
on Microchip PIC32 class devices.

Signed-off-by: Joshua Henderson <joshua.henderson@microchip.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

When (!ctx->bufcnt && !(ctx->flags & SHA_FLAGS_PAD)), the former source
code used to set the SHA_FLAGS_BUSY without checking whether this flag was
already set. If so, the hardware is already processing another hash
request so the processing of the req argument of atmel_sha_final() should
be delayed by queueing this request, the same way as done for the
(ctx->bufcnt != 0) case.

Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Using only the digest, digcnt[], bufcnt and buffer[] fields of the
struct atmel_sha_reqctx was not enough to import/export the request state,
so now we use the whole structure.

Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

(2nd try that adds missing , to build.)

Signed-off-by: Marcus Meissner <meissner@suse.de>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

This needs to go through the security tree so I'm reverting the
patches for now.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Some more authenc() wrapped algorithms are FIPS compliant, tag
them as such.

Signed-off-by: Marcus Meissner <meissner@suse.de>
Acked-by: Stephan Mueller <smueller@chronox.de>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

__test_aead() reads MAX_IVLEN bytes from template[i].iv, but the
actual length of the initialisation vector can be shorter.
The length of the IV is already calculated earlier in the
function. Let's just reuses that. Also the IV length is currently
calculated several time for no reason. Let's fix that too.
This fix an out-of-bound error detected by KASan.

Signed-off-by: Jerome Marchand <jmarchan@redhat.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Currently the sahara driver fails to probe:

sahara: probe of 63ff8000.crypto failed with error -22

This happens since commit 8996eafd

 ("crypto: ahash - ensure statesize
is non-zero"), which requires statesize to be filled.

Pass the statesize members for sha1 and sha256, so we can probe
the driver successfully again.

Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Based on commit 434b4212

 ("crypto: caam - avoid needlessly saving and
restoring caam_hash_ctx") from Russell King.

When exporting and importing the hash state, we will only export and
import into hashes which share the same struct crypto_ahash pointer.
(See hash_accept->af_alg_accept->hash_accept_parent.)

This means that saving the sahara_ctx structure on export, and
restoring it on import is a waste of resources.  So, remove this code.

Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

As pointed out by Herbert Xu we should not include the mutex in the
exported state, so let's just get rid of it.

Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Modify __test_hash() so that hash import/export can be tested
from within the kernel. The test is unconditionally done when
a struct hash_testvec has its .np > 1.

v3: make the test unconditional
v2: Leverage template[i].np as suggested by Tim Chen

Signed-off-by: Rui Wang <rui.y.wang@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

After digsig_asymmetric.c is converted the MPIs can be now
safely removed from the public_key_signature structure.

Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Convert asymmetric_verify to akcipher api.

Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

This patch converts the module verification code to the new akcipher API.

Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Use a local variable for the exported and imported state so that
alignment is not an issue. On export, set a local variable from the
request context and then memcpy the contents of the local variable to
the export memory area. On import, memcpy the import memory area into
a local variable and then use the local variable to set the request
context.

Cc: <stable@vger.kernel.org> # 3.14.x-
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

On  Monday, February 1, 2016 4:18 PM, Herbert Xu wrote:
>
> On Wed, Jan 27, 2016 at 05:08:35PM +0800, Rui Wang wrote:
>>
>> +static int sha1_mb_async_import(struct ahash_request *req, const void
>> +*in) {
>> +	struct ahash_request *mcryptd_req = ahash_request_ctx(req);
>> +	struct crypto_ahash *tfm = crypto_ahash_reqtfm(req);
>> +	struct sha1_mb_ctx *ctx = crypto_ahash_ctx(tfm);
>> +	struct mcryptd_ahash *mcryptd_tfm = ctx->mcryptd_tfm;
>> +	struct crypto_shash *child = mcryptd_ahash_child(mcryptd_tfm);
>> +	struct mcryptd_hash_request_ctx *rctx;
>> +	struct shash_desc *desc;
>> +	int err;
>> +
>> +	memcpy(mcryptd_req, req, sizeof(*req));
>> +	ahash_request_set_tfm(mcryptd_req, &mcryptd_tfm->base);
>> +	rctx = ahash_request_ctx(mcryptd_req);
>> +	desc = &rctx->desc;
>> +	desc->tfm = child;
>> +	desc->flags = CRYPTO_TFM_REQ_MAY_SLEEP;
>> +
>> +	err = crypto_shash_init(desc);
>> +	if (err)
>> +		return err;
>
> What is this desc for?

Hi Herbert,

Yeah I just realized that the call to crypto_shash_init() isn't necessary
here. What it does is overwritten by crypto_ahash_import(). But this desc
still needs to be initialized here because it's newly allocated by
ahash_request_alloc(). We eventually calls the shash version of import()
which needs desc as an argument. The real context to be imported is then
derived from shash_desc_ctx(desc).

desc is a sub-field of struct mcryptd_hash_request_ctx, which is again a
sub-field of the bigger blob allocated by ahash_request_alloc(). The entire
blob's size is set in sha1_mb_async_init_tfm(). So a better version is as
follows:

(just removed the call to crypto_shash_init())

>From 4bcb73adbef99aada94c49f352063619aa24d43d Mon Sep 17 00:00:00 2001
From: Rui Wang <rui.y.wang@intel.com>
Date: Mon, 14 Dec 2015 17:22:13 +0800
Subject: [PATCH v2 1/4] crypto x86/sha1_mb: Fix load failure

modprobe sha1_mb fails with the following message:

modprobe: ERROR: could not insert 'sha1_mb': No such device

It is because it needs to set its statesize and implement its
import() and export() interface.

v2: remove redundant call to crypto_shash_init()

Signed-off-by: Rui Wang <rui.y.wang@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Move the helper function to common header for everybody to use.

changes in v2:
- move the helper to crypto/internal/aead.h
  instead of crypto/aead.h

Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

The driver for the sunxi-ss crypto engine is not entirely 64-bit safe,
compilation on arm64 spits some warnings.
The proper fix was deemed to involved [1], so since 64-bit SoCs won't
have this IP block we just disable this driver for 64-bit.

[1]: http://lists.infradead.org/pipermail/linux-arm-kernel/2016-January/399988.html


     (and the reply)

Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Acked-by: Corentin LABBE <clabbe.montjoie@gmail.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

This patch removes all traces of the crypto_hash interface, now
that everyone has switched over to shash or ahash.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

This patch replaces uses of the long obsolete hash interface with
ahash.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

This patch adds helpers to retrieve the alg name and driver name
of crypto_shash and crypto_ahash objects.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

The helper crypto_skcipher_driver_name was returning the alg
name and not the driver name.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

This patch replaces the crypto_hash example in api-intro.txt with
crypto_ahash.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

This patch removes the last user of the obsolete crypto_hash
interface, tcrypt, by simply switching it over to ahash.  In
fact it already has all the code there so it's just a matter
of calling the ahash speed test code with the right mask.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

We're clearing the wrong memory.  The memory corruption is likely
harmless because we weren't going to use that stack memory again but not
zeroing is a potential information leak.

Fixes: e28facde

 ('crypto: keywrap - add key wrapping block chaining mode')
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Stephan Mueller <smueller@chronox.de>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Since the exported information can be exposed to user-space, instead of
exporting the entire request context only export the minimum information
needed.

Cc: <stable@vger.kernel.org> # 3.14.x-
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Integrate with the newly added crypto engine to make the crypto hardware
engine underutilized as each block needs to be processed before the crypto
hardware can start working on the next block.

The requests from dm-crypt will be listed into engine queue and processed
by engine automatically, so remove the 'queue' and 'queue_task' things in
omap aes driver.

Signed-off-by: Baolin Wang <baolin.wang@linaro.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

Now block cipher engines need to implement and maintain their own queue/thread
for processing requests, moreover currently helpers provided for only the queue
itself (in crypto_enqueue_request() and crypto_dequeue_request()) but they
don't help with the mechanics of driving the hardware (things like running the
request immediately, DMA map it or providing a thread to process the queue in)
even though a lot of that code really shouldn't vary that much from device to
device.

Thus this patch provides a mechanism for pushing requests to the hardware
as it becomes free that drivers could use. And this framework is patterned
on the SPI code and has worked out well there.
(https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/
 drivers/spi/spi.c?id=ffbbdd21

)

Signed-off-by: Baolin Wang <baolin.wang@linaro.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

This patch introduces crypto_queue_len() helper function to help to get the
queue length in the crypto queue list now.

Signed-off-by: Baolin Wang <baolin.wang@linaro.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

This patch implements the missing .import() and .export() mandatory
hooks for asynchronous hash algorithms. It also sets the relevant, non
zero, value for the .statesize field when declaring the supported SHA
algorithms. Indeed a zero value of .statesize prevents the algorithm from
being registered.

Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

The generic crc32 implementation is currently called crc32.  This
is a problem because it clashes with the lib implementation of crc32.

This patch renames the crypto crc32 to crc32_generic so that it is
consistent with crc32c.  An alias for the driver is also added.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

mcryptd_create_hash() fails by returning -EINVAL, causing any
driver using mcryptd to fail to load. It is because it needs
to set its statesize properly.

Signed-off-by: Rui Wang <rui.y.wang@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

When building the jitterentropy driver by itself, we get a link error
when CRYPTO_RNG is not enabled as well:

crypto/built-in.o: In function `jent_mod_init':
jitterentropy-kcapi.c:(.init.text+0x98): undefined reference to `crypto_register_rng'
crypto/built-in.o: In function `jent_mod_exit':
jitterentropy-kcapi.c:(.exit.text+0x60): undefined reference to `crypto_unregister_rng'

This adds a 'select CRYPTO_RNG' to CRYPTO_JITTERENTROPY to ensure the API
is always there when it's used, not just when DRBG is also enabled.
CRYPTO_DRBG would set it implicitly through CRYPTO_JITTERENTROPY now,
but this leaves it in place to make it explicit what the driver does.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

gcc correctly warns that the printk output contains a variable that
it thinks is not initialized in some cases:

drivers/crypto/sunxi-ss/sun4i-ss-cipher.c: In function 'sun4i_ss_cipher_poll':
drivers/crypto/sunxi-ss/sun4i-ss-cipher.c:254:76: warning: 'todo' may be used uninitialized in this function [-Wmaybe-uninitialized]
drivers/crypto/sunxi-ss/sun4i-ss-cipher.c:144:15: note: 'todo' was declared here

A closer look at the function reveals that the variable is always
initialized at this point (ileft is guaranteed to be positive at the
start), but its contents are not well-defined:
Depending on some other variables, it might be either a count in
words or bytes, and it could refer to either input or output.

The easiest solution apparently is to remove the confusing output
and let the reader figure out the state from the other variables.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

This patch removes the unnecessary ivsize variabls as they always
have the value of AES_BLOCK_SIZE.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

It is unused now, so remove it.

Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

The _args_digest is defined as _args+_digest, both of which are the first
members of 2 separate structures, effectively yielding _args_digest to have
a value of zero. Thus, no errors have spawned yet due to this. To ensure
sanity, adding the missing _args_digest offset to the sha1_mb_mgr_submit.S.

Signed-off-by: Megha Dey <megha.dey@linux.intel.com>
Acked-by: Tim Chen <tim.c.chen@linux.intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>