| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Dummy patch to fix typo and indentation.
Signed-off-by: Cyrille Pitchen <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
This is a dummy cosmetic patch.
Signed-off-by: Cyrille Pitchen <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
This patch totally reworks data transfer.
1 - DMA
The new code now fully supports scatter-gather lists hence reducing the
number of interrupts in some cases. Also buffer alignments are better
managed to avoid useless copies.
2 - CPU
The new code allows to use PIO accesses even when transferring more than
one AES block, so futher patches could tune the DMA threshold
(ATMEL_AES_DMA_THRESHOLD).
Moreover, CPU transfers now have a chance to be processed synchronously,
hence reducing the latency by avoiding context switches when possible
(less interrupts to process, less scheduling of the 'done' task).
Indeed the 'DATA READY' bit is polled only one time in the Interrupt
Status Register before enabling then waiting for the associated interrupt.
In some condition, this single poll is enough as the data have already
been processed by the AES hardware and so are ready.
Signed-off-by: Cyrille Pitchen <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Add missing call to atmel_aes_buff_cleanup().
Signed-off-by: Cyrille Pitchen <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Since the 'done' task code was split into atmel_aes_cpu_complete() and
atmel_aes_dma_complete(), the AES_FLAGS_DMA flag has become useless.
Signed-off-by: Cyrille Pitchen <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
atmel_aes_dma_callback() now directly calls the 'resume' callback instead
of scheduling the done task, which in turn only calls the very same
'resume' callback.
Signed-off-by: Cyrille Pitchen <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
This 'err' member was initialized to 0 but its value never changed.
Signed-off-by: Cyrille Pitchen <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
This patch introduces a new callback 'resume' in the struct atmel_aes_dev.
This callback is run to resume/complete the processing of the crypto
request when woken up by I/O events such as AES interrupts or DMA
completion.
This callback will help implementing the GCM mode support in further
patches.
Signed-off-by: Cyrille Pitchen <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
This patch reworks the AES_FLAGS_* to simplify the configuration of the
AES IP.
Signed-off-by: Cyrille Pitchen <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
As claimed by the datasheet, writing 0 into the Control Register has no
effet. So we remove this useless register access.
Signed-off-by: Cyrille Pitchen <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
This patch changes atmel_aes_handle_queue() to make it more generic.
The function argument is now a pointer to struct crypto_async_request,
which is the common base of struct ablkcipher_request and
struct aead_request.
Also this patch introduces struct atmel_aes_base_ctx which will be the
common base of all the transformation contexts.
Hence the very same queue will be used to manage both block cipher and
AEAD requests (such as gcm and authenc implemented in further patches).
Signed-off-by: Cyrille Pitchen <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
This patch changes the signature of atmel_aes_write_ctrl() to make it more
generic. This will be used by future patches when implementing new block
cipher modes such as GCM.
Especially atmel_aes_hw_init() is now called outside
atmel_aes_write_ctrl(): this allows to call atmel_aes_write_ctrl() many
times, still initializing the hardware only once.
Indeed, the support of GCM will require to update the Mode Register and
the IV when processing a single request.
Signed-off-by: Cyrille Pitchen <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Before this patch atmel_aes_hw_version_init() had no returned value.
However it calls atmel_aes_hw_init(), which may fail. So check the
returned code of atmel_aes_hw_init() and propagate error if needed.
Signed-off-by: Cyrille Pitchen <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Hash headers have nothing to do with AES block ciphers.
Signed-off-by: Cyrille Pitchen <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
This dummy patch fixes atmel_aes_unregister_algs() so crypto algorithms
are unregistered in the reverse order they were registered by
atmel_aes_register_algs().
Signed-off-by: Cyrille Pitchen <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Increase the algorithm priorities so the hardware acceleration is now
preferred to the software computation: the "aes-generice" driver uses 100
as priority.
Signed-off-by: Cyrille Pitchen <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
atmel_aes_write_n() should not modify its value argument.
Signed-off-by: Cyrille Pitchen <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Add new version of atmel-aes available with SAMA5D2 devices.
Signed-off-by: Leilei Zhao <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
The atmel AES driver assumes that 'int' and 'size_t' are the same
type in multiple locations, which the compiler warns about when
building it for 64-bit systems:
In file included from ../drivers/crypto/atmel-aes.c:17:0:
drivers/crypto/atmel-aes.c: In function 'atmel_aes_sg_copy':
include/linux/kernel.h:724:17: warning: comparison of distinct pointer types lacks a cast
drivers/crypto/atmel-aes.c:448:11: note: in expansion of macro 'min'
drivers/crypto/atmel-aes.c: In function 'atmel_aes_crypt_dma_stop':
include/linux/kern_levels.h:4:18: warning: format '%u' expects argument of type 'unsigned int', but argument 2 has type 'size_t {aka long unsigned int}' [-Wformat=]
This changes the format strings to use the %z modifier when printing
a size_t, and makes sure that we use the correct size_t type where
needed. In case of sg_dma_len(), the type of the result depends
on CONFIG_NEED_SG_DMA_LENGTH, so we have to use min_t to get it to
work in all configurations.
Signed-off-by: Arnd Bergmann <[email protected]>
Acked-by: Cyrille Pitchen <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Using the devm_xxx() managed function to stripdown the error and remove
code.
Signed-off-by: LABBE Corentin <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
clk_prepare_enable() can fail so add a check for this and
return the error code if it fails.
Signed-off-by: LABBE Corentin <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
The output buffer is used for CPU access, so
the API should be dma_sync_single_for_cpu which
makes the cache line invalid in order to reload
the value in memory.
Signed-off-by: Leilei Zhao <[email protected]>
Acked-by: Nicolas Ferre <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
The input buffer and output buffer are mapped for DMA transfer
in Atmel AES driver. But they are also be used by CPU when
the requested crypt length is not bigger than the threshold
value 16. The buffers will be cached in cache line when CPU
accessed them. When DMA uses the buffers again, the memory
can happened to be flushed by cache while DMA starts transfer.
So using API dma_sync_single_for_device and dma_sync_single_for_cpu
in DMA to ensure DMA coherence and CPU always access the correct
value. This fix the issue that the encrypted result periodically goes
wrong when doing performance test with OpenSSH.
Signed-off-by: Leilei Zhao <[email protected]>
Acked-by: Nicolas Ferre <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Kernel will report "BUG: spinlock lockup suspected on CPU#0"
when CONFIG_DEBUG_SPINLOCK is enabled in kernel config and the
spinlock is used at the first time. It's caused by uninitialized
spinlock, so just initialize it in probe.
Signed-off-by: Leilei Zhao <[email protected]>
Acked-by: Nicolas Ferre <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Add new version of atmel-aes available with SAMA5D4 devices.
Signed-off-by: Leilei Zhao <[email protected]>
Signed-off-by: Ludovic Desroches <[email protected]>
Acked-by: Nicolas Ferre <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Fix typo, "intialization" -> "initialization"
Signed-off-by: Colin Ian King <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
If only one of the 2 __get_free_pages fails, then there is a memory leak.
Signed-off-by: Christophe Jaillet <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
A platform_driver does not need to set an owner, it will be populated by the
driver core.
Signed-off-by: Wolfram Sang <[email protected]>
|
|
The length shoule be 64 bit alignment and the block size shoule be 64 bit in aes cfb64 mode.
Signed-off-by: Leilei Zhao <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
The block size of aes cfb8 mode shoule be 8 bit.
Signed-off-by: Leilei Zhao <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Add support for Device Tree and use of the DMA DT API to
get the needed channels.
Documentation is added for these DT nodes.
Initial code by: Nicolas Royer and Eukrea.
Signed-off-by: Nicolas Ferre <[email protected]>
Acked-by: Herbert Xu <[email protected]>
|
|
Updates from previous IP release (0x120):
- add cfb64 support
- add DMA double input buffer support
Signed-off-by: Nicolas Royer <[email protected]>
Acked-by: Nicolas Ferre <[email protected]>
Acked-by: Eric Bénard <[email protected]>
Tested-by: Eric Bénard <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Pull crypto update from Herbert Xu:
"Here is the crypto update for 3.9:
- Added accelerated implementation of crc32 using pclmulqdq.
- Added test vector for fcrypt.
- Added support for OMAP4/AM33XX cipher and hash.
- Fixed loose crypto_user input checks.
- Misc fixes"
* git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6: (43 commits)
crypto: user - ensure user supplied strings are nul-terminated
crypto: user - fix empty string test in report API
crypto: user - fix info leaks in report API
crypto: caam - Added property fsl,sec-era in SEC4.0 device tree binding.
crypto: use ERR_CAST
crypto: atmel-aes - adjust duplicate test
crypto: crc32-pclmul - Kill warning on x86-32
crypto: x86/twofish - assembler clean-ups: use ENTRY/ENDPROC, localize jump labels
crypto: x86/sha1 - assembler clean-ups: use ENTRY/ENDPROC
crypto: x86/serpent - use ENTRY/ENDPROC for assember functions and localize jump targets
crypto: x86/salsa20 - assembler cleanup, use ENTRY/ENDPROC for assember functions and rename ECRYPT_* to salsa20_*
crypto: x86/ghash - assembler clean-up: use ENDPROC at end of assember functions
crypto: x86/crc32c - assembler clean-up: use ENTRY/ENDPROC
crypto: cast6-avx: use ENTRY()/ENDPROC() for assembler functions
crypto: cast5-avx: use ENTRY()/ENDPROC() for assembler functions and localize jump targets
crypto: camellia-x86_64/aes-ni: use ENTRY()/ENDPROC() for assembler functions and localize jump targets
crypto: blowfish-x86_64: use ENTRY()/ENDPROC() for assembler functions and localize jump targets
crypto: aesni-intel - add ENDPROC statements for assembler functions
crypto: x86/aes - assembler clean-ups: use ENTRY/ENDPROC, localize jump targets
crypto: testmgr - add test vector for fcrypt
...
|
|
Delete successive tests to the same location. The code tested the result
of a previous allocation, that itself was already tested. It is changed to
test the result of the most recent allocation.
A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)
// <smpl>
@s exists@
local idexpression y;
expression x,e;
@@
*if ( \(x == NULL\|IS_ERR(x)\|y != 0\) )
{ ... when forall
return ...; }
... when != \(y = e\|y += e\|y -= e\|y |= e\|y &= e\|y++\|y--\|&y\)
when != \(XT_GETPAGE(...,y)\|WMI_CMD_BUF(...)\)
*if ( \(x == NULL\|IS_ERR(x)\|y != 0\) )
{ ... when forall
return ...; }
// </smpl>
Signed-off-by: Julia Lawall <[email protected]>
Acked-by: David S. Miller <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
CONFIG_HOTPLUG is going away as an option. As a result, the __dev*
markings need to be removed.
This change removes the use of __devinit, __devexit_p, __devinitdata,
and __devexit from these drivers.
Based on patches originally written by Bill Pemberton, but redone by me
in order to handle some of the coding style issues better, by hand.
Cc: Bill Pemberton <[email protected]>
Cc: Herbert Xu <[email protected]>
Cc: "David S. Miller" <[email protected]>
Cc: Kent Yoder <[email protected]>
Cc: Jamie Iles <[email protected]>
Cc: Kim Phillips <[email protected]>
Cc: Shengzhou Liu <[email protected]>
Cc: Alex Porosanu <[email protected]>
Signed-off-by: Greg Kroah-Hartman <[email protected]>
|
|
From: Wei Yongjun <[email protected]>
Remove duplicated include.
Signed-off-by: Wei Yongjun <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Initialization of cra_list is currently mixed, most ciphers initialize this
field and most shashes do not. Initialization however is not needed at all
since cra_list is initialized/overwritten in __crypto_register_alg() with
list_add(). Therefore perform cleanup to remove all unneeded initializations
of this field in 'crypto/drivers/'.
Cc: Benjamin Herrenschmidt <[email protected]>
Cc: [email protected]
Cc: Michal Ludvig <[email protected]>
Cc: Dmitry Kasatkin <[email protected]>
Cc: Varun Wadekar <[email protected]>
Cc: Eric Bénard <[email protected]>
Signed-off-by: Jussi Kivilinna <[email protected]>
Acked-by: Kent Yoder <[email protected]>
Acked-by: Vladimir Zapolskiy <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Signed-off-by: Nicolas Royer <[email protected]>
Acked-by: Nicolas Ferre <[email protected]>
Acked-by: Eric Bénard <[email protected]>
Tested-by: Eric Bénard <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|