It could be a situation, that tasklet is executed twice because of
certain delay between dma callback and irq handler execution.
In that case, second tasklet execution could actually corrupt the data
of the new started dma transactions.
This patch improves tasklet logic and prevents above described cases.
Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Calling omap_sham_handle_queue from "done" tasklet should be done
after irq scheduled tasklet completes.
Having additional tasklet does not solve that issue because it might
be execute before.
So queue handling tasklet has been removed and functionality integrated
into single tasklet.
Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Couple of context flags have been moved to device flags.
IRQ and tasklet handlers does not need to access request
context anymore.
Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Some flags are changed in interrupt handlers and verified in the tasklet.
There might be a race condition when tasklet is interrupted or another
cpu/core will run IRQ handler and tasklet in parallel.
Atomic bitops functions are now used instead of bitmask operations.
Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Flags mask cannot be used with atomic bit operations.
This patch changes masks to bit numbers.
Atomic bit operations will be used by following patches.
Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Request pointer is already available in the function.
Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
This patch fixes 2 hmac inter-dependent bugs.
1. "omap-sham: hash-in-progress is stored in hw format" commit introduced
optimization where temporary hash had been stored in OMAP specific format
(big endian).
For SHA1 it is different to real hash format, which is little endian.
Final HMAC value was calculated using incorrect hash.
Because CONFIG_CRYPTO_MANAGER_TESTS was disabled this error remained
unnoticed. After enabling this option, bug has been found.
2. HMAC was calculated using temporrary hash value.
For a single-request updates, temporary hash was the final one and
HMAC result was correct. But in fact only the final hash had to be used.
All crypto tests for HMAC produces only single request and
could not catch the problem. This problem is fixed here.
Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
EMU chips also have crypto HW as HS chips.
Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
SHA1 accelerator can also be busy. Add -EBUSY status return option and
return busy status from omap_sham_finup().
Signed-off-by: Markku Kylanpaa <ext-markku.kylanpaa@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
clk_get() returns a struct clk cookie to the driver and some platforms
may return NULL if they only support a single clock. clk_get() has only
failed if it returns a ERR_PTR() encoded pointer.
Signed-off-by: Jamie Iles <jamie@jamieiles.com>
Reviewed-by: Aaro Koskinen <aaro.koskinen@nokia.com>
Reviewed-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Previous commit "removed redundant locking" introduced
a bug in handling backlog.
In certain cases, when async request complete callback will
call complete() on -EINPROGRESS code, it will cause uncompleted requests.
It does not happen in implementation similar to crypto test manager,
but it will happen in implementation similar to dm-crypt.
Backlog needs to be checked before dequeuing next request.
Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
If scatterlist have more than one entry, current driver uses
aligned buffer to copy data to to accelerator to tackle possible
issues with DMA and SHA buffer alignment.
This commit adds more intelligence to verify SG alignment and
possibility to use DMA directly on the data without using copy
buffer.
Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
bufcnt is 0 if it was no update requests before,
which is exact meaning of FLAGS_FIRST.
Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Hash-in-progress is now stored in hw format.
Only on final call, hash is converted to correct format.
Speedup copy procedure and will allow to use OMAP burst mode.
Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
According to the Herbert Xu, client may not always call
crypto_ahash_final().
In the case of error in hash calculation resources will be
automatically cleaned up.
But if no hash calculation error happens and client will not call
crypto_ahash_final() at all, then internal buffer will not be freed,
and clocks will not be disabled.
This patch provides support for atomic crypto_ahash_update() call.
Clocks are now enabled and disabled per update request.
Data buffer is now allocated as a part of request context.
Client is obligated to free it with crypto_free_ahash().
Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Locking for queuing and dequeuing is combined.
test_and_set_bit() is also replaced with checking under dd->lock.
Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Introduces DMA error handling.
DMA error is returned as a result code of the hash request.
Clients needs to handle error codes and may repeat hash calculation attempt.
Also in the case of DMA error, SHAM module is set to be re-initialized again.
It significantly improves stability against possible HW failures.
Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
DMA parameters for constant data were initialized during driver probe().
It seems that those settings sometimes are lost when devices goes to off mode.
This patch makes DMA initialization just before use.
It solves off mode problems.
Fixes: NB#202786 - Aegis & SHA1 block off mode changes
Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Currently driver storred digest results in req->results
provided by the client. But some clients do not set it
until final() call. It leads to crash.
Changed to use internal buffer to store temporary digest results.
Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
DMA is set to use burst mode also for source channel. It should
descrease memory bandwidth needs.
DMA synchronization is set to use prefetch mechanism. SHAM block is behind
L4 bus and it doesn't have fifo. SHAM block is stalling as long as the new
data is available. It takes time to fetch data from memory and transfer it
via L4 bus. With prefetch enabled, data is waiting in DMA fifo and SHAM block
receives new data block faster.
This increases SHA processing speed up to 30 percent depending on the
bus / memory load.
Signed-off-by: Samu Onkalo <samu.p.onkalo@nokia.com>
Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Earlier kernel contained omap sha1 and md5 driver, which was not maintained,
was not ported to new crypto APIs and removed from the source tree.
- implements async crypto API using dma and cpu.
- supports multiple sham instances if available
- hmac
- concurrent requests
Signed-off-by: Dmitry Kasatkin <dmitry.kasatkin@nokia.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Dmitry Kasatkin