- 30 May, 2019 20 commits
-
-
Christophe Leroy authored
The driver assumes that the ICV is as a single piece in the last element of the scatterlist. This assumption is wrong. This patch ensures that the ICV is properly handled regardless of the scatterlist layout. Fixes: 9c4a7965 ("crypto: talitos - Freescale integrated security engine (SEC) driver") Signed-off-by:
Christophe Leroy <christophe.leroy@c-s.fr> Signed-off-by:
Herbert Xu <herbert@gondor.apana.org.au>
-
Christophe Leroy authored
In that mode, hardware ICV verification is not supported. Signed-off-by:
-
Christophe Leroy authored
For decrypt, req->cryptlen includes the size of the authentication part while all functions of the driver expect cryptlen to be the size of the encrypted data. As it is not expected to change req->cryptlen, this patch implements local calculation of cryptlen. Signed-off-by:
-
Christophe Leroy authored
ECB's ivsize must be 0. Signed-off-by:
-
Christophe Leroy authored
When data size is not a multiple of the alg's block size, the SEC generates an error interrupt and dumps the registers. And for NULL size, the SEC does just nothing and the interrupt is awaited forever. This patch ensures the data size is correct before submitting the request to the SEC engine. Signed-off-by:
-
Christophe Leroy authored
CTR has a blocksize of 1. Signed-off-by:
-
Christophe Leroy authored
Although the HW accepts any size and silently truncates it to the correct length, the extra tests expects EINVAL to be returned when the key size is not valid. Signed-off-by:
-
Christophe Leroy authored
SEC1 doesn't support SHA384/512, so it doesn't require longer keys. This patch reduces the max key size when the driver is built for SEC1 only. Signed-off-by:
-
Christophe Leroy authored
The talitos driver has two ways to perform AEAD depending on the HW capability. Some HW support both. It is needed to give them different names to distingish which one it is for instance when a test fails. Signed-off-by:
-
Eric Biggers authored
Remove the crypto_tfm_in_queue() function, which is unused. Signed-off-by:
Eric Biggers <ebiggers@google.com> Signed-off-by:
-
Eric Biggers authored
Remove the unnecessary constant CRYPTO_ALG_TYPE_DIGEST, which has the same value as CRYPTO_ALG_TYPE_HASH. Signed-off-by:
-
Eric Biggers authored
kcrypto_wq is only used by cryptd, so move it into cryptd.c and change the workqueue name from "crypto" to "cryptd". Signed-off-by:
-
Eric Biggers authored
There's no reason for users to select CONFIG_CRYPTO_GF128MUL, since it's just some helper functions, and algorithms that need it select it. Remove the prompt string so that it's not shown to users. Signed-off-by:
-
Eric Biggers authored
echainiv is the only algorithm or template in the crypto API that is enabled by default. But there doesn't seem to be a good reason for it. And it pulls in a lot of stuff as dependencies, like AEAD support and a "NIST SP800-90A DRBG" including HMAC-SHA256. The commit which made it default 'm', commit 3491244c ("crypto: echainiv - Set Kconfig default to m"), mentioned that it's needed for IPsec. However, later commit 32b6170c ("ipv4+ipv6: Make INET*_ESP select CRYPTO_ECHAINIV") made the IPsec kconfig options select it. So, remove the 'default m'. Signed-off-by:
-
Eric Biggers authored
The "cryptomgr" module is required for templates to be used. Many templates select it, but others don't. Make all templates select it. Signed-off-by:
-
Eric Biggers authored
The crypto self-tests are part of the "cryptomgr" module, which can technically be disabled (though it rarely is). If you do so, currently you can still enable CRYPTO_MANAGER_EXTRA_TESTS, which doesn't make sense since in that case testmgr.c isn't compiled at all. Fix it by making it CRYPTO_MANAGER_EXTRA_TESTS depend on CRYPTO_MANAGER2, like CRYPTO_MANAGER_DISABLE_TESTS already does. Fixes: 5b2706a4 ("crypto: testmgr - introduce CONFIG_CRYPTO_MANAGER_EXTRA_TESTS") Signed-off-by:
-
Eric Biggers authored
On PowerPC with CONFIG_CRYPTO_MANAGER_EXTRA_TESTS=y, there is sometimes a crash in generate_random_aead_testvec(). The problem is that the generated test vectors use data lengths of up to about 2 * PAGE_SIZE, which is 128 KiB on PowerPC; however, the data length fields in the test vectors are 'unsigned short', so the lengths get truncated. Fix this by changing the relevant fields to 'unsigned int'. Fixes: 40153b10 ("crypto: testmgr - fuzz AEADs against their generic implementation") Signed-off-by:
-
Eric Biggers authored
Convert the VMX implementations of AES-CBC, AES-CTR, and AES-XTS from the deprecated "blkcipher" API to the "skcipher" API. As part of this, I moved the skcipher_request for the fallback algorithm off the stack and into the request context of the parent algorithm. I tested this in a PowerPC VM with CONFIG_CRYPTO_MANAGER_EXTRA_TESTS=y. Signed-off-by:
Michael Ellerman <mpe@ellerman.id.au> Signed-off-by:
-
Eric Biggers authored
Remove the boilerplate license text and replace it with the equivalent SPDX license identifier. Signed-off-by:
-
Neil Armstrong authored
Signed-off-by:
Neil Armstrong <narmstrong@baylibre.com> Signed-off-by:
-
- 23 May, 2019 20 commits
-
-
Christian Lamparter authored
The hardware automatically zero pads incomplete block ciphers blocks without raising any errors. This is a screw-up. This was noticed by CONFIG_CRYPTO_MANAGER_EXTRA_TESTS tests that sent a incomplete blocks and expect them to fail. This fixes: cbc-aes-ppc4xx encryption unexpectedly succeeded on test vector "random: len=2409 klen=32"; expected_error=-22, cfg="random: may_sleep use_digest src_divs=[96.90%@+2295, 2.34%@+4066, 0.32%@alignmask+12, 0.34%@+4087, 0.9%@alignmask+1787, 0.1%@+3767] iv_offset=6" ecb-aes-ppc4xx encryption unexpectedly succeeded on test vector "random: len=1011 klen=32"; expected_error=-22, cfg="random: may_sleep use_digest src_divs=[100.0%@alignmask+20] dst_divs=[3.12%@+3001, 96.88%@+4070]" Cc: Eric Biggers <ebiggers@kernel.org> Cc: stable@vger.kernel.org [4.19, 5.0 and 5.1] Signed-off-by:
Christian Lamparter <chunkeey@gmail.com> Signed-off-by:
-
Christian Lamparter authored
While the hardware consider them to be blockciphers, the reference implementation defines them as streamciphers. Do the right thing and set the blocksize to 1. This was found by CONFIG_CRYPTO_MANAGER_EXTRA_TESTS. This fixes the following issues: skcipher: blocksize for ofb-aes-ppc4xx (16) doesn't match generic impl (1) skcipher: blocksize for cfb-aes-ppc4xx (16) doesn't match generic impl (1) Cc: Eric Biggers <ebiggers@kernel.org> Cc: stable@vger.kernel.org Fixes: f2a13e7c ("crypto: crypto4xx - enable AES RFC3686, ECB, CFB and OFB offloads") Signed-off-by:
-
Christian Lamparter authored
This patch fixes a issue with crypto4xx's ctr(aes) that was discovered by libcapi's kcapi-enc-test.sh test. The some of the ctr(aes) encryptions test were failing on the non-power-of-two test: kcapi-enc - Error: encryption failed with error 0 kcapi-enc - Error: decryption failed with error 0 [FAILED: 32-bit - 5.1.0-rc1+] 15 bytes: STDIN / STDOUT enc test (128 bits): original file (1d100e..cc96184c) and generated file (e3b0c442..1b7852b855) [FAILED: 32-bit - 5.1.0-rc1+] 15 bytes: STDIN / STDOUT enc test (128 bits) (openssl generated CT): original file (e3b0..5) and generated file (3..8e) [PASSED: 32-bit - 5.1.0-rc1+] 15 bytes: STDIN / STDOUT enc test (128 bits) (openssl generated PT) [FAILED: 32-bit - 5.1.0-rc1+] 15 bytes: STDIN / STDOUT enc test (password): original file (1d1..84c) and generated file (e3b..852b855) But the 16, 32, 512, 65536 tests always worked. Thankfully, this isn't a hidden hardware problem like previously, instead this turned out to be a copy and paste issue. With this patch, all the tests are passing with and kcapi-enc-test.sh gives crypto4xx's a clean bill of health: "Number of failures: 0" :). Cc: stable@vger.kernel.org Fixes: 98e87e3d ("crypto: crypto4xx - add aes-ctr support") Fixes: f2a13e7c ("crypto: crypto4xx - enable AES RFC3686, ECB, CFB and OFB offloads") Signed-off-by:
-
Sascha Hauer authored
For encryption the destination pointer was still mapped, so the hex dump may be wrong. The IV still contained the input IV while printing instead of the output IV as intended. For decryption the destination pointer was still mapped, so the hex dump may be wrong. The IV dump was correct. Do the hex dumps consistenly after the buffers have been unmapped and in case of IV copied to their final destination. Signed-off-by:
Sascha Hauer <s.hauer@pengutronix.de> Reviewed-by:
Horia Geantă <horia.geanta@nxp.com> Signed-off-by:
-
Christophe Leroy authored
Selftests report the following: [ 2.984845] alg: skcipher: cbc-aes-talitos encryption test failed (wrong output IV) on test vector 0, cfg="in-place" [ 2.995377] 00000000: 3d af ba 42 9d 9e b4 30 b4 22 da 80 2c 9f ac 41 [ 3.032673] alg: skcipher: cbc-des-talitos encryption test failed (wrong output IV) on test vector 0, cfg="in-place" [ 3.043185] 00000000: fe dc ba 98 76 54 32 10 [ 3.063238] alg: skcipher: cbc-3des-talitos encryption test failed (wrong output IV) on test vector 0, cfg="in-place" [ 3.073818] 00000000: 7d 33 88 93 0f 93 b2 42 This above dumps show that the actual output IV is indeed the input IV. This is due to the IV not being copied back into the request. This patch fixes that. Signed-off-by:
-
Hook, Gary authored
Crypto self-tests reveal an error: alg: skcipher: cbc-des3-ccp encryption test failed (wrong output IV) on test vector 0, cfg="in-place" The offset value should not be recomputed when retrieving the context. Also, a code path exists which makes decisions based on older (version 3) hardware; a v3 device deosn't support 3DES so remove this check. Fixes: 990672d4 ('crypto: ccp - Enable 3DES function on v5 CCPs') Signed-off-by:
Gary R Hook <gary.hook@amd.com> Signed-off-by:
-
Hook, Gary authored
Updated testmgr will exhibit this error message when loading the ccp-crypto module: alg: skcipher: cfb-aes-ccp encryption failed with err -22 on test vector 3, cfg="in-place" Update the CCP crypto driver to correctly treat CFB as a streaming mode cipher (instead of block mode). Update the configuration for CFB to specify the block size as a single byte; Fixes: 2b789435 ('crypto: ccp - CCP AES crypto API support') Signed-off-by:
-
Hook, Gary authored
CFB mode should be treated as a stream cipher, not block. Fixes: 63b94509 ('crypto: ccp - CCP device driver and interface support') Signed-off-by:
-
YueHaibing authored
Fix sparse warning: arch/arm/crypto/sha512-glue.c:40:5: warning: symbol 'sha512_arm_final' was not declared. Should it be static? Reported-by:
Hulk Robot <hulkci@huawei.com> Signed-off-by:
YueHaibing <yuehaibing@huawei.com> Signed-off-by:
-
Stephan Mueller authored
FIPS 140-2 section 4.9.2 requires a continuous self test of the noise source. Up to kernel 4.8 drivers/char/random.c provided this continuous self test. Afterwards it was moved to a location that is inconsistent with the FIPS 140-2 requirements. The relevant patch was e192be9d . Thus, the FIPS 140-2 CTRNG is added to the DRBG when it obtains the seed. This patch resurrects the function drbg_fips_continous_test that existed some time ago and applies it to the noise sources. The patch that removed the drbg_fips_continous_test was b3614763 . The Jitter RNG implements its own FIPS 140-2 self test and thus does not need to be subjected to the test in the DRBG. The patch contains a tiny fix to ensure proper zeroization in case of an error during the Jitter RNG data gathering. Signed-off-by:
Stephan Mueller <smueller@chronox.de> Reviewed-by:
Yann Droneaud <ydroneaud@opteya.com> Signed-off-by:
-
Horia Geantă authored
Currently there is a mismatch b/w the ICID (Isolation Context ID) used for DMA mapping keys and ICID used for accessing them. -keys are DMA mapped using a job ring device, thus a job ring ICID -keys are accessed from descriptors enqueued via Queue Interface, thus using QI ICID [Note: ICIDs of JRs, QI are configured by U-boot / other entity by: -fixing up the corresponding job ring and controller DT nodes -setting up corresponding caam ICID registers] In order to avoid IOMMU faults, DMA map the key using the controller device instead of a job ring device. Signed-off-by:
-
Horia Geantă authored
When IOMMU is enabled, iova -> phys address translation should be performed using iommu_ops, not dma_to_phys(). Signed-off-by:
-
Horia Geantă authored
Use the controller device for caam/qi instead of allocating a new platform device. This is needed as a preparation to add support for working behind an SMMU. A platform device allocated using platform_device_register_full() is not completely set up - most importantly .dma_configure() is not called. Signed-off-by:
-
Horia Geantă authored
Currently we allow top level code, i.e. that which sits between the low level (HW-specific) drivers and crypto API, to be built as several drivers: caamalg, caamhash, caam_pkc, caamrng, caamalg_qi. There is no advantage in this, more it interferes with adding support for deferred probing (there are no corresponding devices and thus no bus). Convert these drivers and call init() / exit() manually at the right time. Move algorithms initialization at JR probe / remove time: -the first probed JR registers the crypto algs -the last removed JR unregisters the crypto algs Note: caam_qi_init() is called before JR platform devices creation (of_populate_bus()), such that QI interface is initialized when the caam/qi algorithms are registered in the JR driver (by calling caam_qi_algapi_init(). While here, fix the Kconfig entries under CRYPTO_DEV_FSL_CAAM_JR to be aligned. Signed-off-by:
-
Horia Geantă authored
According to CAAM RM: -crypto engine reads 4 S/G entries (64 bytes) at a time, even if the S/G table has fewer entries -it's the responsibility of the user / programmer to make sure this HW behaviour has no side effect The drivers do not take care of this currently, leading to IOMMU faults when the S/G table ends close to a page boundary - since only one page is DMA mapped, while CAAM's DMA engine accesses two pages. Fix this by rounding up the number of allocated S/G table entries to a multiple of 4. Note that in case of two *contiguous* S/G tables, only the last table might needs extra entries. Signed-off-by:
-
Horia Geantă authored
When enabling IOMMU support, the following issue becomes visible in the AEAD zero-length case. Even though the output sequence length is set to zero, the crypto engine tries to prefetch 4 S/G table entries (since SGF bit is set in SEQ OUT PTR command - which is either generated in SW in case of caam/jr or in HW in case of caam/qi, caam/qi2). The DMA read operation will trigger an IOMMU fault since the address in the SEQ OUT PTR is "dummy" (set to zero / not obtained via DMA API mapping). 1. In case of caam/jr, avoid the IOMMU fault by clearing the SGF bit in SEQ OUT PTR command. 2. In case of caam/qi - setting address, bpid, length to zero for output entry in the compound frame has a special meaning (cf. CAAM RM): "Output frame = Unspecified, Input address = Y. A unspecified frame is indicated by an unused SGT entry (an entry in which the Address, Length, and BPID fields are all zero). SEC obtains output buffers from BMan as prescribed by the preheader." Since no output buffers are needed, modify the preheader by setting (ABS = 1, ADDBUF = 0): -"ABS = 1 means obtain the number of buffers in ADDBUF (0 or 1) from the pool POOL ID" -ADDBUF: "If ABS is set, ADD BUF specifies whether to allocate a buffer or not" 3. In case of caam/qi2, since engine: -does not support FLE[FMT]=2'b11 ("unused" entry) mentioned in DPAA2 RM -requires output entry to be present, even if not used the solution chosen is to leave output frame list entry zeroized. Fixes: 763069ba ("crypto: caam - handle zero-length AEAD output") Signed-off-by: -
Giovanni Cabiddu authored
If a zero length request is submitted through the skcipher api, do not offload it and return success. Reviewed-by:
Conor Mcloughlin <conor.mcloughlin@intel.com> Tested-by:
Sergey Portnoy <sergey.portnoy@intel.com> Signed-off-by:
Giovanni Cabiddu <giovanni.cabiddu@intel.com> Signed-off-by:
-
Giovanni Cabiddu authored
Return -EINVAL if a request for a block cipher is not multiple of the size of the block. This problem was found with by the new extra run-time crypto self test. Reviewed-by:
-
Giovanni Cabiddu authored
If an invalid key is provided as input to the setkey function, the function always failed returning -ENOMEM rather than -EINVAL. Furthermore, if setkey was called multiple times with an invalid key, the device instance was getting leaked. This patch fixes the error paths in the setkey functions by returning the correct error code in case of error and freeing all the resources allocated in this function in case of failure. This problem was found with by the new extra run-time crypto self test. Reviewed-by:
-
Giovanni Cabiddu authored
The block size for aes counter mode was improperly set to AES_BLOCK_SIZE. This sets it to 1 as it is a stream cipher. This problem was found with by the new extra run-time crypto self test. Reviewed-by:
-
