]> git.karo-electronics.de Git - linux-beck.git/log
linux-beck.git
9 years agocrypto: caam - pair irq map and dispose in the same function
Cristian Stoica [Thu, 22 Jan 2015 14:00:48 +0000 (16:00 +0200)]
crypto: caam - pair irq map and dispose in the same function

irq_dispose_mapping is not called on all error paths from caam_jr_init.
This takes care of several clean-up issues by performing resource
clean-up and allocation at the same level.

Signed-off-by: Cristian Stoica <cristian.stoica@freescale.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: ccp - terminate ccp_support array with empty element
Andrey Ryabinin [Wed, 21 Jan 2015 15:06:31 +0000 (18:06 +0300)]
crypto: ccp - terminate ccp_support array with empty element

x86_match_cpu() expects array of x86_cpu_ids terminated
with empty element.

Signed-off-by: Andrey Ryabinin <a.ryabinin@samsung.com>
Acked-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: caam - remove unused local variable
Cristian Stoica [Wed, 21 Jan 2015 09:53:31 +0000 (11:53 +0200)]
crypto: caam - remove unused local variable

Signed-off-by: Cristian Stoica <cristian.stoica@freescale.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: caam - remove dead code
Cristian Stoica [Wed, 21 Jan 2015 09:53:30 +0000 (11:53 +0200)]
crypto: caam - remove dead code

- assoc_nents and src_nents are never zero when all_contig is false
- iv_contig is zero on the else branch

Signed-off-by: Cristian Stoica <cristian.stoica@freescale.com>
Reviewed-by: Richard Schmitt <richard.schmitt@freescale.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: caam - don't emit ICV check failures to dmesg
Kim Phillips [Tue, 20 Jan 2015 18:43:10 +0000 (12:43 -0600)]
crypto: caam - don't emit ICV check failures to dmesg

ICV check failures are part of normal operation;
leave user notification up to the higher levels,
as is done in s/w algorithm implementations.

Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
Tested-by: Cristian Stoica <cristian.stoica@freescale.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agohwrng: virtio - drop extra empty line
Michael S. Tsirkin [Tue, 20 Jan 2015 12:35:15 +0000 (14:35 +0200)]
hwrng: virtio - drop extra empty line

makes code look a bit prettier.

Cc: linux-crypto@vger.kernel.org.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: replace scatterwalk_sg_next with sg_next
Cristian Stoica [Tue, 20 Jan 2015 08:06:16 +0000 (10:06 +0200)]
crypto: replace scatterwalk_sg_next with sg_next

Modify crypto drivers to use the generic SG helper since
both of them are equivalent and the one from crypto is redundant.

See also:
  468577abe37ff7b453a9ac613e0ea155349203ae   reverted in
  b2ab4a57b018aafbba35bff088218f5cc3d2142e

Signed-off-by: Cristian Stoica <cristian.stoica@freescale.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: atmel - Free memory in error path
Christophe Jaillet [Tue, 20 Jan 2015 07:15:52 +0000 (08:15 +0100)]
crypto: atmel - Free memory in error path

If only one of the 2 __get_free_pages fails, then there is a memory leak.

Signed-off-by: Christophe Jaillet <christophe.jaillet@wanadoo.fr>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: doc - remove colons in comments
Stephan Mueller [Sun, 18 Jan 2015 23:13:39 +0000 (00:13 +0100)]
crypto: doc - remove colons in comments

As documented in Documentation/kernel-doc-nano-HOWTO.txt lines
terminated with a colon are treated as headings.

The current layout of the documentation when compiling the kernel
crypto API DocBook documentation is messed up by by treating some lines
as headings. The patch removes colons from comments that shall not be
treated as headings.

Signed-off-by: Stephan Mueller <smueller@chronox.de>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: seqiv - Ensure that IV size is at least 8 bytes
Herbert Xu [Fri, 16 Jan 2015 08:51:20 +0000 (19:51 +1100)]
crypto: seqiv - Ensure that IV size is at least 8 bytes

Since seqiv is designed for IPsec we need to be able to accomodate
the whole IPsec sequence number in order to ensure the uniqueness
of the IV.

This patch forbids any algorithm with an IV size of less than 8
from using it.  This should have no impact on existing users since
they all have an IV size of 8.

Reported-by: Maciej ?enczykowski <zenczykowski@gmail.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Acked-by: Maciej ?enczykowski <zenczykowski@gmail.com>
9 years agocrypto: cts - Weed out non-CBC algorithms
Herbert Xu [Fri, 16 Jan 2015 08:38:17 +0000 (19:38 +1100)]
crypto: cts - Weed out non-CBC algorithms

The cts algorithm as currently implemented assumes the underlying
is a CBC-mode algorithm.  So this patch adds a check for that to
eliminate bogus combinations of cts with non-CBC modes.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agoMAINTAINERS: add linux-crypto to hw random
Michael S. Tsirkin [Fri, 16 Jan 2015 07:16:00 +0000 (09:16 +0200)]
MAINTAINERS: add linux-crypto to hw random

hw random is crypto-related, Cc the linux-crypto list
on patches.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: cts - Remove bogus use of seqiv
Herbert Xu [Fri, 16 Jan 2015 07:09:21 +0000 (18:09 +1100)]
crypto: cts - Remove bogus use of seqiv

The seqiv generator is completely inappropriate for cts as it's
designed for IPsec algorithms.  Since cts users do not actually
use the IV generator we can just fall back to the default.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Acked-by: Maciej ?enczykowski <zenczykowski@gmail.com>
9 years agocrypto: qat - don't need qat_auth_state struct
Tadeusz Struk [Wed, 14 Jan 2015 18:27:35 +0000 (10:27 -0800)]
crypto: qat - don't need qat_auth_state struct

We don't need the qat_auth_state structure anymore.

Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: algif_rng - fix sparse non static symbol warning
Wei Yongjun [Wed, 14 Jan 2015 01:14:41 +0000 (09:14 +0800)]
crypto: algif_rng - fix sparse non static symbol warning

Fixes the following sparse warnings:

crypto/algif_rng.c:185:13: warning:
 symbol 'rng_exit' was not declared. Should it be static?

Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Acked-by: Stephan Mueller <smueller@chronox.de>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: qat - Ensure ipad and opad are zeroed
Herbert Xu [Tue, 13 Jan 2015 23:07:03 +0000 (10:07 +1100)]
crypto: qat - Ensure ipad and opad are zeroed

The patch ad511e260a27b8e35d273cc0ecfe5a8ff9543181 (crypto: qat -
Fix incorrect uses of memzero_explicit) broke hashing because the
code was in fact overwriting the qat_auth_state variable.

In fact there is no reason for the variable to exist anyway since
all we are using it for is to store ipad and opad.  So we could
simply create ipad and opad directly and avoid this whole mess.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: aesni - Add support for 192 & 256 bit keys to AESNI RFC4106
Timothy McCaffrey [Tue, 13 Jan 2015 18:16:43 +0000 (13:16 -0500)]
crypto: aesni - Add support for 192 & 256 bit keys to AESNI RFC4106

These patches fix the RFC4106 implementation in the aesni-intel
module so it supports 192 & 256 bit keys.

Since the AVX support that was added to this module also only
supports 128 bit keys, and this patch only affects the SSE
implementation, changes were also made to use the SSE version
if key sizes other than 128 are specified.

RFC4106 specifies that 192 & 256 bit keys must be supported (section
8.4).

Also, this should fix Strongswan issue 341 where the aesni module
needs to be unloaded if 256 bit keys are used:

http://wiki.strongswan.org/issues/341

This patch has been tested with Sandy Bridge and Haswell processors.
With 128 bit keys and input buffers > 512 bytes a slight performance
degradation was noticed (~1%).  For input buffers of less than 512
bytes there was no performance impact.  Compared to 128 bit keys,
256 bit key size performance is approx. .5 cycles per byte slower
on Sandy Bridge, and .37 cycles per byte slower on Haswell (vs.
SSE code).

This patch has also been tested with StrongSwan IPSec connections
where it worked correctly.

I created this diff from a git clone of crypto-2.6.git.

Any questions, please feel free to contact me.

Signed-off-by: Timothy McCaffrey <timothy.mccaffrey@unisys.com>
Signed-off-by: Jarod Wilson <jarod@redhat.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: x86/des3_ede - drop bogus module aliases
Mathias Krause [Sun, 11 Jan 2015 17:17:47 +0000 (18:17 +0100)]
crypto: x86/des3_ede - drop bogus module aliases

This module implements variations of "des3_ede" only. Drop the bogus
module aliases for "des".

Cc: Jussi Kivilinna <jussi.kivilinna@iki.fi>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: sparc64/md5 - fix module description
Mathias Krause [Sun, 11 Jan 2015 17:17:46 +0000 (18:17 +0100)]
crypto: sparc64/md5 - fix module description

MD5 is not SHA1.

Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: sparc64/des - add "des3_ede" module alias
Mathias Krause [Sun, 11 Jan 2015 17:17:45 +0000 (18:17 +0100)]
crypto: sparc64/des - add "des3_ede" module alias

This module provides implementations for "des3_ede", too. Announce those
via an appropriate crypto module alias so it can be used in favour to
the generic C implementation.

Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: sparc64/camellia - fix module alias
Mathias Krause [Sun, 11 Jan 2015 17:17:44 +0000 (18:17 +0100)]
crypto: sparc64/camellia - fix module alias

The module alias should be "camellia", not "aes".

Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: sparc64/aes - fix module description
Mathias Krause [Sun, 11 Jan 2015 17:17:43 +0000 (18:17 +0100)]
crypto: sparc64/aes - fix module description

AES is a block cipher, not a hash.

Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: qat - remove unnecessary include of atomic.h header file
Allan, Bruce W [Fri, 9 Jan 2015 19:55:20 +0000 (11:55 -0800)]
crypto: qat - remove unnecessary include of atomic.h header file

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: qat - use pci_wait_for_pending_transaction()
Allan, Bruce W [Fri, 9 Jan 2015 19:55:14 +0000 (11:55 -0800)]
crypto: qat - use pci_wait_for_pending_transaction()

Prior to resetting the hardware, use pci_wait_for_pending_transaction()
instead of open coding similar functionality.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Acked-by: Tadeusz Struk <tadeusz.struk@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: qat - adf_ae_stop() is never called
Allan, Bruce W [Fri, 9 Jan 2015 19:55:09 +0000 (11:55 -0800)]
crypto: qat - adf_ae_stop() is never called

In adf_dev_stop(), adf_ae_stop() is never called because adf_dev_started()
will always return false since the ADF_STATUS_STARTED bit is cleared
earlier in the function.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Acked-by: Tadeusz Struk <tadeusz.struk@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: qat - correctly type a boolean
Allan, Bruce W [Fri, 9 Jan 2015 19:55:04 +0000 (11:55 -0800)]
crypto: qat - correctly type a boolean

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: qat - fix device reset flow
Allan, Bruce W [Fri, 9 Jan 2015 19:54:58 +0000 (11:54 -0800)]
crypto: qat - fix device reset flow

When the device needs a reset, e.g. when an uncorrectable PCIe AER event
occurs, various services/data structures need to be cleaned up, the
hardware reset and the services/data structures initialized and started.
The code to perform the cleanup and initialization was not performed when
a device reset was done.

This patch moves some of the initialization code out of the .probe entry-
point into a separate function that is now called during probe as well as
after the hardware has been reset.  Similarly, a new function is added for
first cleaning up these services/data structures prior to resetting.  The
new functions are adf_dev_init() and adf_dev_shutdown(), respectively, for
which there are already prototypes but no actual functions just yet and are
now called when the device is reset and during probe/cleanup of the driver.
The down and up flows via ioctl calls has similarly been updated.

In addition, there are two other bugs in the reset flow - one in the logic
for determining whether to schedule a device reset upon receiving an
uncorrectable AER event which prevents the reset flow from being initiated,
and another with clearing the status bit indicating a device is configured
(when resetting the device the configuration remains across the reset so
the bit should not be cleared, otherwise, the necessary services will not
be re-started in adf_dev_start() after the reset - clear the bit only when
actually deleting the configuration).

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: testmgr - don't use interruptible wait in tests
Rabin Vincent [Fri, 9 Jan 2015 15:25:28 +0000 (16:25 +0100)]
crypto: testmgr - don't use interruptible wait in tests

tcrypt/testmgr uses wait_for_completion_interruptible() everywhere when
it waits for a request to be completed.  If it's interrupted, then the
test is aborted and the request is freed.

However, if any of these calls actually do get interrupted, the result
will likely be a kernel crash, when the driver handles the now-freed
request.  Use wait_for_completion() instead.

Signed-off-by: Rabin Vincent <rabin.vincent@axis.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agolib: memzero_explicit: add comment for its usage
Daniel Borkmann [Mon, 5 Jan 2015 23:27:45 +0000 (00:27 +0100)]
lib: memzero_explicit: add comment for its usage

Lets improve the comment to add a note on when to use memzero_explicit()
for those not digging through the git logs. We don't want people to
pollute places with memzero_explicit() where it's not really necessary.

Reference: https://lkml.org/lkml/2015/1/4/190
Suggested-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: qat - Fix incorrect uses of memzero_explicit
Herbert Xu [Mon, 5 Jan 2015 20:54:41 +0000 (07:54 +1100)]
crypto: qat - Fix incorrect uses of memzero_explicit

memzero_explicit should only be used on stack variables that get
zapped just before they go out of scope.

This patch replaces all unnecessary uses of memzero_explicit with
memset, removes two memzero_explicit calls altogether as the tfm
context comes pre-zeroed, and adds a missing memzero_explicit of
the stack variable buff in qat_alg_do_precomputes.  The memzeros
on ipad/opad + digest_size/auth_keylen are also removed as the
entire auth_state is already zeroed on entry.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Acked-by: Tadeusz Struk <tadeusz.struk@intel.com>
9 years agocrypto: aead - add check for presence of auth tag
Stephan Mueller [Mon, 5 Jan 2015 11:21:45 +0000 (12:21 +0100)]
crypto: aead - add check for presence of auth tag

The AEAD decryption operation requires the authentication tag to be
present as part of the cipher text buffer. The added check verifies that
the caller provides a cipher text with at least the authentication tag.

Signed-off-by: Stephan Mueller <smueller@chronox.de>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: omap-des - fix BUG_ON condition
Asaf Vertz [Mon, 5 Jan 2015 08:23:10 +0000 (10:23 +0200)]
crypto: omap-des - fix BUG_ON condition

dd->total is unsigned so it won't do any good to check for negative
value after subtracting instead of that we should check if the
subtracted value is bigger than him

This was partially found by using a static code analysis program
called cppcheck.

Signed-off-by: Asaf Vertz <asaf.vertz@tandemg.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: bfin_crc - Remove unnecessary KERN_ERR in bfin_crc.c
Masanari Iida [Fri, 2 Jan 2015 03:40:46 +0000 (12:40 +0900)]
crypto: bfin_crc - Remove unnecessary KERN_ERR in bfin_crc.c

This patch removes unnecessary KERN_ERR from bfin_crypto_crc_mod_init().

Signed-off-by: Masanari Iida <standby24x7@gmail.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: amcc - Remove unused function
Rickard Strandqvist [Thu, 1 Jan 2015 15:13:47 +0000 (16:13 +0100)]
crypto: amcc - Remove unused function

Remove the function get_dynamic_sa_offset_iv_field() that is not used anywhere.

This was partially found by using a static code analysis program called cppcheck.

Signed-off-by: Rickard Strandqvist <rickard_strandqvist@spectrumdigital.se>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agoRevert "crypto: drbg - use memzero_explicit() for clearing sensitive data"
Herbert Xu [Sun, 4 Jan 2015 23:44:09 +0000 (10:44 +1100)]
Revert "crypto: drbg - use memzero_explicit() for clearing sensitive data"

This reverts commit 421d82f5b3e75f94e31875e37d45cdf6a557c120.

None of the data zeroed are on the stack so the compiler cannot
optimise them away.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: algif_rng - enable RNG interface compilation
Stephan Mueller [Thu, 25 Dec 2014 22:00:39 +0000 (23:00 +0100)]
crypto: algif_rng - enable RNG interface compilation

Enable compilation of the RNG AF_ALG support and provide a Kconfig
option to compile the RNG AF_ALG support.

Signed-off-by: Stephan Mueller <smueller@chronox.de>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: algif_rng - add random number generator support
Stephan Mueller [Thu, 25 Dec 2014 22:00:06 +0000 (23:00 +0100)]
crypto: algif_rng - add random number generator support

This patch adds the random number generator support for AF_ALG.

A random number generator's purpose is to generate data without
requiring the caller to provide any data. Therefore, the AF_ALG
interface handler for RNGs only implements a callback handler for
recvmsg.

The following parameters provided with a recvmsg are processed by the
RNG callback handler:

* sock - to resolve the RNG context data structure accessing the
  RNG instance private to the socket

* len - this parameter allows userspace callers to specify how
  many random bytes the RNG shall produce and return. As the
  kernel context for the RNG allocates a buffer of 128 bytes to
  store random numbers before copying them to userspace, the len
  parameter is checked that it is not larger than 128. If a
  caller wants more random numbers, a new request for recvmsg
  shall be made.

The size of 128 bytes is chose because of the following considerations:

* to increase the memory footprint of the kernel too much (note,
  that would be 128 bytes per open socket)

* 128 is divisible by any typical cryptographic block size an
  RNG may have

* A request for random numbers typically only shall supply small
  amount of data like for keys or IVs that should only require
  one invocation of the recvmsg function.

Note, during instantiation of the RNG, the code checks whether the RNG
implementation requires seeding. If so, the RNG is seeded with output
from get_random_bytes.

A fully working example using all aspects of the RNG interface is
provided at http://www.chronox.de/libkcapi.html

Signed-off-by: Stephan Mueller <smueller@chronox.de>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: af_alg - zeroize key data
Stephan Mueller [Tue, 23 Dec 2014 08:34:03 +0000 (09:34 +0100)]
crypto: af_alg - zeroize key data

alg_setkey should zeroize the sensitive data after use.

Signed-off-by: Stephan Mueller <smueller@chronox.de>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agohwrng: core - Move hwrng_init call into set_current_rng
Herbert Xu [Tue, 23 Dec 2014 05:40:22 +0000 (16:40 +1100)]
hwrng: core - Move hwrng_init call into set_current_rng

We always do hwrng_init in set_current_rng.  In fact, our current
reference count system relies on this.  So make this explicit by
moving hwrng_init into set_current_rng.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agohwrng: core - Drop current rng in set_current_rng
Herbert Xu [Tue, 23 Dec 2014 05:40:21 +0000 (16:40 +1100)]
hwrng: core - Drop current rng in set_current_rng

Rather than having callers of set_current_rng call drop_current_rng,
we can do it directly in set_current_rng.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agohwrng: core - Do not register device opportunistically
Herbert Xu [Tue, 23 Dec 2014 05:40:19 +0000 (16:40 +1100)]
hwrng: core - Do not register device opportunistically

Currently we only register the device when a valid RNG is added.
However the way it's done is buggy because we test whether there
is a current RNG to determine whether we need to register.  As
the current RNG may be missing due to a reinitialisation error
this can lead to a reregistration of the device.

As the device already has to handle a NULL current RNG anyway,
let's just register the device always and remove the complexity.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agohwrng: core - Fix current_rng init/cleanup race yet again
Herbert Xu [Tue, 23 Dec 2014 05:40:18 +0000 (16:40 +1100)]
hwrng: core - Fix current_rng init/cleanup race yet again

The kref solution is still buggy because we were only focusing
on the register/unregister race.  The same race affects the
setting of current_rng through sysfs.

This patch fixes it by using kref_get_unless_zero.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agohwrng: core - Use struct completion for cleanup_done
Herbert Xu [Tue, 23 Dec 2014 05:40:17 +0000 (16:40 +1100)]
hwrng: core - Use struct completion for cleanup_done

There is no point in doing a manual completion for cleanup_done
when struct completion fits in perfectly.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: octeon - enable OCTEON MD5 module selection
Aaro Koskinen [Sun, 21 Dec 2014 20:54:02 +0000 (22:54 +0200)]
crypto: octeon - enable OCTEON MD5 module selection

Enable user to select OCTEON MD5 module.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: octeon - add MD5 module
Aaro Koskinen [Sun, 21 Dec 2014 20:54:01 +0000 (22:54 +0200)]
crypto: octeon - add MD5 module

Add OCTEON MD5 module.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agoMIPS: OCTEON: reintroduce crypto features check
Aaro Koskinen [Sun, 21 Dec 2014 20:54:00 +0000 (22:54 +0200)]
MIPS: OCTEON: reintroduce crypto features check

Reintroduce run-time check for crypto features. The old one was deleted
because it was unreliable, now decide the crypto availability on early
boot when the model string is constructed.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: octeon - add instruction definitions for MD5
Aaro Koskinen [Sun, 21 Dec 2014 20:53:59 +0000 (22:53 +0200)]
crypto: octeon - add instruction definitions for MD5

Add instruction definitions for MD5. Based on information extracted
from EdgeRouter Pro GPL source tarball.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agoMIPS: OCTEON: add crypto helper functions
Aaro Koskinen [Sun, 21 Dec 2014 20:53:58 +0000 (22:53 +0200)]
MIPS: OCTEON: add crypto helper functions

Add crypto helper functions which are needed for kernel level usage.
The code for these has been extracted from the EdgeRouter Pro GPL tarball.

While at it, also delete duplicate definitions of the functions.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: ux500 - fix checkpatch errors
Asaf Vertz [Wed, 10 Dec 2014 12:55:10 +0000 (14:55 +0200)]
crypto: ux500 - fix checkpatch errors

Fixed a coding style error, code indent should use tabs where possible

Signed-off-by: Asaf Vertz <asaf.vertz@tandemg.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: qat - add support for cbc(aes) ablkcipher
Tadeusz Struk [Mon, 8 Dec 2014 20:08:49 +0000 (12:08 -0800)]
crypto: qat - add support for cbc(aes) ablkcipher

Add support for cbc(aes) ablkcipher.

Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
Acked-by: Bruce W. Allan <bruce.w.allan@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: qat - Fix assumption that sg in and out will have the same nents
Tadeusz Struk [Mon, 8 Dec 2014 20:05:42 +0000 (12:05 -0800)]
crypto: qat - Fix assumption that sg in and out will have the same nents

Fixed invalid assumpion that the sgl in and sgl out will always have the same
number of entries.

Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: algif - Mark sgl end at the end of data
Tadeusz Struk [Mon, 8 Dec 2014 20:03:42 +0000 (12:03 -0800)]
crypto: algif - Mark sgl end at the end of data

algif_skcipher sends 127 sgl buffers for encryption regardless of how
many buffers acctually have data to process, where the few first with
valid len and the rest with zero len. This is not very eficient.
This patch marks the last one with data as the last one to process.

Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agohwrng: don't init list element we're about to add to list.
Rusty Russell [Mon, 8 Dec 2014 08:50:40 +0000 (16:50 +0800)]
hwrng: don't init list element we're about to add to list.

Another interesting anti-pattern.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agohwrng: don't double-check old_rng.
Rusty Russell [Mon, 8 Dec 2014 08:50:39 +0000 (16:50 +0800)]
hwrng: don't double-check old_rng.

Interesting anti-pattern.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agohwrng: fix unregister race.
Rusty Russell [Mon, 8 Dec 2014 08:50:38 +0000 (16:50 +0800)]
hwrng: fix unregister race.

The previous patch added one potential problem: we can still be
reading from a hwrng when it's unregistered.  Add a wait for zero
in the hwrng_unregister path.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Amos Kong <akong@redhat.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agohwrng: use reference counts on each struct hwrng.
Rusty Russell [Mon, 8 Dec 2014 08:50:37 +0000 (16:50 +0800)]
hwrng: use reference counts on each struct hwrng.

current_rng holds one reference, and we bump it every time we want
to do a read from it.

This means we only hold the rng_mutex to grab or drop a reference,
so accessing /sys/devices/virtual/misc/hw_random/rng_current doesn't
block on read of /dev/hwrng.

Using a kref is overkill (we're always under the rng_mutex), but
a standard pattern.

This also solves the problem that the hwrng_fillfn thread was
accessing current_rng without a lock, which could change (eg. to NULL)
underneath it.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Amos Kong <akong@redhat.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agohwrng: move some code out mutex_lock for avoiding underlying deadlock
Amos Kong [Mon, 8 Dec 2014 08:50:36 +0000 (16:50 +0800)]
hwrng: move some code out mutex_lock for avoiding underlying deadlock

In next patch, we use reference counting for each struct hwrng,
changing reference count also needs to take mutex_lock. Before
releasing the lock, if we try to stop a kthread that waits to
take the lock to reduce the referencing count, deadlock will
occur.

Signed-off-by: Amos Kong <akong@redhat.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agohwrng: place mutex around read functions and buffers.
Rusty Russell [Mon, 8 Dec 2014 08:50:35 +0000 (16:50 +0800)]
hwrng: place mutex around read functions and buffers.

There's currently a big lock around everything, and it means that we
can't query sysfs (eg /sys/devices/virtual/misc/hw_random/rng_current)
while the rng is reading.  This is a real problem when the rng is slow,
or blocked (eg. virtio_rng with qemu's default /dev/random backend)

This doesn't help (it leaves the current lock untouched), just adds a
lock to protect the read function and the static buffers, in preparation
for transition.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: af_alg - add setsockopt for auth tag size
Stephan Mueller [Sun, 7 Dec 2014 22:21:42 +0000 (23:21 +0100)]
crypto: af_alg - add setsockopt for auth tag size

Use setsockopt on the tfm FD to provide the authentication tag size for
an AEAD cipher. This is achieved by adding a callback function which is
intended to be used by the AEAD AF_ALG implementation.

The optlen argument of the setsockopt specifies the authentication tag
size to be used with the AEAD tfm.

Signed-off-by: Stephan Mueller <smueller@chronox.de>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: drbg - panic on continuous self test error
Stephan Mueller [Fri, 5 Dec 2014 21:40:21 +0000 (22:40 +0100)]
crypto: drbg - panic on continuous self test error

This patch adds a panic if the FIPS 140-2 self test error failed.
Note, that entire code is only executed with fips_enabled (i.e. when the
kernel is booted with fips=1. It is therefore not executed for 99.9% of
all user base.

As mathematically such failure cannot occur, this panic should never be
triggered. But to comply with NISTs current requirements, an endless
loop must be replaced with the panic.

When the new version of FIPS 140 will be released, this entire
continuous self test function will be ripped out as it will not be
needed any more.

This patch is functionally equivalent as implemented in ansi_cprng.c and drivers/char/random.c.

Signed-off-by: Stephan Mueller <smueller@chronox.de>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: api - fixed style erro in algapi.c
Joshua I. James [Fri, 5 Dec 2014 06:00:10 +0000 (15:00 +0900)]
crypto: api - fixed style erro in algapi.c

Fixed style error identified by checkpatch.

WARNING: Missing a blank line after declarations
+               int err = crypto_remove_alg(&inst->alg, &users);
+               BUG_ON(err);

Signed-off-by: Joshua I. James <joshua@cybercrimetech.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: ahash - fixed style error in ahash.c
Joshua I. James [Fri, 5 Dec 2014 05:44:54 +0000 (14:44 +0900)]
crypto: ahash - fixed style error in ahash.c

Fixed style error identified by checkpatch.

WARNING: Missing a blank line after declarations
+               unsigned int unaligned = alignmask + 1 - (offset & alignmask);
+               if (nbytes > unaligned)

Signed-off-by: Joshua I. James <joshua@cybercrimetech.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: af_alg - fixed style error in af_alg.c
Joshua I. James [Fri, 5 Dec 2014 05:38:40 +0000 (14:38 +0900)]
crypto: af_alg - fixed style error in af_alg.c

Fixed style error identified by checkpatch.

ERROR: space required before the open parenthesis '('
+               switch(cmsg->cmsg_type) {

Signed-off-by: Joshua I. James <joshua@cybercrimetech.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: aead - fixed style error in aead.c
Joshua I. James [Fri, 5 Dec 2014 05:24:44 +0000 (14:24 +0900)]
crypto: aead - fixed style error in aead.c

Fixed style error identified by checkpatch.

ERROR: do not use assignment in if condition
+       if ((err = crypto_register_instance(tmpl, inst))) {

Signed-off-by: Joshua I. James <joshua@cybercrimetech.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: ablkcipher - fixed style errors in ablkcipher.c
Joshua I. James [Fri, 5 Dec 2014 05:06:16 +0000 (14:06 +0900)]
crypto: ablkcipher - fixed style errors in ablkcipher.c

Fixed style errors reported by checkpatch.

WARNING: Missing a blank line after declarations
+       u8 *end_page = (u8 *)(((unsigned long)(start + len - 1)) & PAGE_MASK);
+       return max(start, end_page);

WARNING: line over 80 characters
+               scatterwalk_start(&walk->out, scatterwalk_sg_next(walk->out.sg));

WARNING: Missing a blank line after declarations
+               int err = ablkcipher_copy_iv(walk, tfm, alignmask);
+               if (err)

ERROR: do not use assignment in if condition
+       if ((err = crypto_register_instance(tmpl, inst))) {

Signed-off-by: Joshua I. James <joshua@cybercrimetech.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: atmel_sha - remove unused shash fallback instance.
Svenning Sørensen [Fri, 5 Dec 2014 00:18:57 +0000 (01:18 +0100)]
crypto: atmel_sha - remove unused shash fallback instance.

The fallback is never used, so there is no point in having it.

The cra_exit routine can also be removed, since all it did was releasing
the fallback, along with the stub around cra_init, which just added an
unused NULL argument.

Signed-off-by: Svenning Soerensen <sss@secomea.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agocrypto: af_alg - fix backlog handling
Rabin Vincent [Fri, 19 Dec 2014 12:36:08 +0000 (13:36 +0100)]
crypto: af_alg - fix backlog handling

If a request is backlogged, it's complete() handler will get called
twice: once with -EINPROGRESS, and once with the final error code.

af_alg's complete handler, unlike other users, does not handle the
-EINPROGRESS but instead always completes the completion that recvmsg()
is waiting on.  This can lead to a return to user space while the
request is still pending in the driver.  If userspace closes the sockets
before the requests are handled by the driver, this will lead to
use-after-frees (and potential crashes) in the kernel due to the tfm
having been freed.

The crashes can be easily reproduced (for example) by reducing the max
queue length in cryptod.c and running the following (from
http://www.chronox.de/libkcapi.html) on AES-NI capable hardware:

 $ while true; do kcapi -x 1 -e -c '__ecb-aes-aesni' \
    -k 00000000000000000000000000000000 \
    -p 00000000000000000000000000000000 >/dev/null & done

Cc: stable@vger.kernel.org
Signed-off-by: Rabin Vincent <rabin.vincent@axis.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
9 years agoLinux 3.19-rc1
Linus Torvalds [Sun, 21 Dec 2014 01:08:50 +0000 (17:08 -0800)]
Linux 3.19-rc1

9 years agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/borntraeger...
Linus Torvalds [Sun, 21 Dec 2014 00:48:59 +0000 (16:48 -0800)]
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/borntraeger/linux

Pull ACCESS_ONCE cleanup preparation from Christian Borntraeger:
 "kernel: Provide READ_ONCE and ASSIGN_ONCE

  As discussed on LKML http://marc.info/?i=54611D86.4040306%40de.ibm.com
  ACCESS_ONCE might fail with specific compilers for non-scalar
  accesses.

  Here is a set of patches to tackle that problem.

  The first patch introduce READ_ONCE and ASSIGN_ONCE.  If the data
  structure is larger than the machine word size memcpy is used and a
  warning is emitted.  The next patches fix up several in-tree users of
  ACCESS_ONCE on non-scalar types.

  This does not yet contain a patch that forces ACCESS_ONCE to work only
  on scalar types.  This is targetted for the next merge window as Linux
  next already contains new offenders regarding ACCESS_ONCE vs.
  non-scalar types"

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/borntraeger/linux:
  s390/kvm: REPLACE barrier fixup with READ_ONCE
  arm/spinlock: Replace ACCESS_ONCE with READ_ONCE
  arm64/spinlock: Replace ACCESS_ONCE READ_ONCE
  mips/gup: Replace ACCESS_ONCE with READ_ONCE
  x86/gup: Replace ACCESS_ONCE with READ_ONCE
  x86/spinlock: Replace ACCESS_ONCE with READ_ONCE
  mm: replace ACCESS_ONCE with READ_ONCE or barriers
  kernel: Provide READ_ONCE and ASSIGN_ONCE

9 years agoMerge tag 'clk-for-linus-3.19' of git://git.linaro.org/people/mike.turquette/linux
Linus Torvalds [Sun, 21 Dec 2014 00:42:36 +0000 (16:42 -0800)]
Merge tag 'clk-for-linus-3.19' of git://git.linaro.org/people/mike.turquette/linux

Pull clk framework updates from Mike Turquette:
 "This is much later than usual due to several last minute bugs that had
  to be addressed.  As usual the majority of changes are new drivers and
  modifications to existing drivers.  The core recieved many fixes along
  with the groundwork for several large changes coming in the future
  which will better parition clock providers from clock consumers"

* tag 'clk-for-linus-3.19' of git://git.linaro.org/people/mike.turquette/linux: (86 commits)
  clk: samsung: Fix Exynos 5420 pinctrl setup and clock disable failure due to domain being gated
  ARM: OMAP3: clock: fix boot breakage in legacy mode
  ARM: OMAP2+: clock: fix DPLL code to use new determine rate APIs
  clk: Really fix deadlock with mmap_sem
  clk: mmp: fix sparse non static symbol warning
  clk: Change clk_ops->determine_rate to return a clk_hw as the best parent
  clk: change clk_debugfs_add_file to take a struct clk_hw
  clk: Don't expose __clk_get_accuracy
  clk: Don't try to use a struct clk* after it could have been freed
  clk: Remove unused function __clk_get_prepare_count
  clk: samsung: Fix double add of syscore ops after driver rebind
  clk: samsung: exynos4: set parent of sclk_hdmiphy to hdmi
  clk: samsung: exynos4415: Fix build with PM_SLEEP disabled
  clk: samsung: remove unnecessary inclusion of header files from clk.h
  clk: samsung: remove unnecessary CONFIG_OF from clk.c
  clk: samsung: Spelling s/bwtween/between/
  clk: rockchip: Add support for the mmc clock phases using the framework
  clk: rockchip: add bindings for the mmc clocks
  clk: rockchip: rk3288 export i2s0_clkout for use in DT
  clk: rockchip: use clock ID for DMC (memory controller) on rk3288
  ...

9 years agoMerge branch 'i2c/for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa...
Linus Torvalds [Sat, 20 Dec 2014 21:52:52 +0000 (13:52 -0800)]
Merge branch 'i2c/for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux

Pull more i2c updates from Wolfram Sang:
 "Included are two bugfixes needing some bigger refactoring (sh_mobile:
  deferred probe with DMA, mv64xxx: fix offload support) and one
  deprecated driver removal I thought would go in via ppc but I
  misunderstood.  It has a proper ack from BenH"

* 'i2c/for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  i2c: sh_mobile: fix uninitialized var when debug is enabled
  macintosh: therm_pm72: delete deprecated driver
  i2c: sh_mobile: I2C_SH_MOBILE should depend on HAS_DMA
  i2c: sh_mobile: rework deferred probing
  i2c: sh_mobile: refactor DMA setup
  i2c: mv64xxx: rework offload support to fix several problems
  i2c: mv64xxx: use BIT() macro for register value definitions

9 years agoMerge tag 'scsi-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb...
Linus Torvalds [Sat, 20 Dec 2014 21:42:57 +0000 (13:42 -0800)]
Merge tag 'scsi-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi

Pull SCSI update from James Bottomley:
 "This is a much shorter set of patches that were on the go but didn't
  make it in to the early pull request for the merge window.  It's
  really a set of bug fixes plus some final cleanup work on the new tag
  queue API"

* tag 'scsi-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  storvsc: ring buffer failures may result in I/O freeze
  ipr: set scsi_level correctly for disk arrays
  ipr: add support for async scanning to speed up boot
  scsi_debug: fix missing "break;" in SDEBUG_UA_CAPACITY_CHANGED case
  scsi_debug: take sdebug_host_list_lock when changing capacity
  scsi_debug: improve driver description in Kconfig
  scsi_debug: fix compare and write errors
  qla2xxx: fix race in handling rport deletion during recovery causes panic
  scsi: blacklist RSOC for Microsoft iSCSI target devices
  scsi: fix random memory corruption with scsi-mq + T10 PI
  Revert "[SCSI] mpt3sas: Remove phys on topology change"
  Revert "[SCSI] mpt2sas: Remove phys on topology change."
  esas2r: Correct typos of "validate" in a comment
  fc: FCP_PTA_SIMPLE is 0
  ibmvfc: remove unused tag variable
  scsi: remove MSG_*_TAG defines
  scsi: remove scsi_set_tag_type
  scsi: remove scsi_get_tag_type
  scsi: never drop to untagged mode during queue ramp down
  scsi: remove ->change_queue_type method

9 years agoMerge tag 'pm-config-3.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafae...
Linus Torvalds [Sat, 20 Dec 2014 21:37:44 +0000 (13:37 -0800)]
Merge tag 'pm-config-3.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Pull CONFIG_PM_RUNTIME elimination from Rafael Wysocki:
 "This removes the last few uses of CONFIG_PM_RUNTIME introduced
  recently and makes that config option finally go away.

  CONFIG_PM will be available directly from the menu now and also it
  will be selected automatically if CONFIG_SUSPEND or CONFIG_HIBERNATION
  is set"

* tag 'pm-config-3.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  PM: Eliminate CONFIG_PM_RUNTIME
  tty: 8250_omap: Replace CONFIG_PM_RUNTIME with CONFIG_PM
  sound: sst-haswell-pcm: Replace CONFIG_PM_RUNTIME with CONFIG_PM
  spi: Replace CONFIG_PM_RUNTIME with CONFIG_PM

9 years agoMerge branch 'misc' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild
Linus Torvalds [Sat, 20 Dec 2014 21:33:03 +0000 (13:33 -0800)]
Merge branch 'misc' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild

Pull misc kbuild changes from Michal Marek:
 "There are only a few things in the misc branch:

   - Fix for bugon.cocci semantic patch
   - Kdevelop4 files are .gitignored
   - Put make binrpm-pkg on diet"

* 'misc' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild:
  scripts/package: binrpm-pkg do not create source and devel package
  .gitignore: Add Kdevelop4 project files
  bugon.cocci: fix Options at the macro

9 years agoMerge branch 'kbuild' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild
Linus Torvalds [Sat, 20 Dec 2014 21:31:14 +0000 (13:31 -0800)]
Merge branch 'kbuild' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild

Pull kbuild updates from Michal Marek:
 "Here are the kbuild changes for v3.19-rc1:

   - Cleanups and deduplication in the main Makefile and
     scripts/Makefile.*
   - Sort the output of *config targets in make help
   - Old <linux/version.h> is always removed to avoid a surprise during
     bisecting
   - Warning fix in kconfig"

* 'kbuild' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild:
  kbuild: remove redundant -rR flag of hdr-inst
  kbuild: Fix make help-<board series> on powerpc
  kbuild: Automatically remove stale <linux/version.h> file
  kconfig: Fix warning "‘jump’ may be used uninitialized"
  Makefile: sort list of defconfig targets in make help output
  kbuild: Remove duplicate $(cmd) definition in Makefile.clean
  kbuild: collect shorthands into scripts/Kbuild.include

9 years agoi2c: sh_mobile: fix uninitialized var when debug is enabled
Wolfram Sang [Tue, 16 Dec 2014 12:31:26 +0000 (13:31 +0100)]
i2c: sh_mobile: fix uninitialized var when debug is enabled

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Sat, 20 Dec 2014 02:19:19 +0000 (18:19 -0800)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs

Pull vfs pile #3 from Al Viro:
 "Assorted fixes and patches from the last cycle"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  [regression] chunk lost from bd9b51
  vfs: make mounts and mountstats honor root dir like mountinfo does
  vfs: cleanup show_mountinfo
  init: fix read-write root mount
  unfuck binfmt_misc.c (broken by commit e6084d4)
  vm_area_operations: kill ->migrate()
  new helper: iter_is_iovec()
  move_extent_per_page(): get rid of unused w_flags
  lustre: get rid of playing with ->fs
  btrfs: filp_open() returns ERR_PTR() on failure, not NULL...

9 years agoMerge tag 'ecryptfs-3.19-rc1-fixes' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 20 Dec 2014 02:15:12 +0000 (18:15 -0800)]
Merge tag 'ecryptfs-3.19-rc1-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tyhicks/ecryptfs

Pull eCryptfs fixes from Tyler Hicks:
 "Fixes for filename decryption and encrypted view plus a cleanup

   - The filename decryption routines were, at times, writing a zero
     byte one character past the end of the filename buffer

   - The encrypted view feature attempted, and failed, to roll its own
     form of enforcing a read-only mount instead of letting the VFS
     enforce it"

* tag 'ecryptfs-3.19-rc1-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tyhicks/ecryptfs:
  eCryptfs: Remove buggy and unnecessary write in file name decode routine
  eCryptfs: Remove unnecessary casts when parsing packet lengths
  eCryptfs: Force RO mount when encrypted view is enabled

9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux...
Linus Torvalds [Sat, 20 Dec 2014 02:10:42 +0000 (18:10 -0800)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs

Pull more btrfs updates from Chris Mason:
 "This is part two of our merge window patches.

  These are all from Filipe, and fix some really hard to find races that
  can cause corruptions.  Most of them involved block group removal
  (balance) or discard"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
  Btrfs: remove non-sense btrfs_error_discard_extent() function
  Btrfs: fix fs corruption on transaction abort if device supports discard
  Btrfs: always clear a block group node when removing it from the tree
  Btrfs: ensure deletion from pinned_chunks list is protected

9 years agoMerge tag 'sound-fix-3.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai...
Linus Torvalds [Sat, 20 Dec 2014 02:07:17 +0000 (18:07 -0800)]
Merge tag 'sound-fix-3.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound

Pull sound fixes from Takashi Iwai:
 "Here are a few fixes that have landed after the previous pull request.
  All are driver specific fixes including:

   - error/int value fixes in OXFW,
   - Intel Skylake HD-audio HDMI codec support,
   - Additional HD-audio Realtek codecs and AD1986A codec fixes/quirks,
   - a few more DSD support and a quirk for Arcam rPAC in usb-audio,
   - a typo fix for Scarlett 6i6,
   - fixes for new ASIHPI firmware,
   - ASoC Exynos7 cleanups,
   - Intel ACPI support, and
   - a fix for PCM512 register cache sync"

* tag 'sound-fix-3.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound: (24 commits)
  ALSA: usb-audio: extend KEF X300A FU 10 tweak to Arcam rPAC
  ALSA: hda/realtek - New codec support for ALC298
  ALSA: asihpi: update to HPI version 4.14
  ALSA: asihpi: increase tuner pad cache size
  ALSA: asihpi: relax firmware version check
  ALSA: usb-audio: Fix Scarlett 6i6 initialization typo
  ALSA: hda - Add quirk for Packard Bell EasyNote MX65
  ALSA: usb-audio: add native DSD support for Matrix Audio DACs
  ALSA: hda/realtek - New codec support for ALC256
  ALSA: hda/realtek - Add new Dell desktop for ALC3234 headset mode
  ASoC: Intel: fix possible acpi enumeration panic
  ALSA: hda/hdmi - apply Haswell fix-ups to Skylake display codec
  ASoC: Intel: fix return value check in sst_acpi_probe()
  ALSA: hda - Make add_stereo_mix_input flag tristate
  ALSA: hda - Create capture source ctls when stereo mix input is added
  ALSA: hda - Fix typos in snd_hda_get_int_hint() kerneldoc comments
  ALSA: hda - add codec ID for Skylake display audio codec
  ALSA: oxfw: some signedness bugs
  ALSA: oxfw: fix detect_loud_models() return value
  ASoC: rt5677: add REGMAP_I2C and REGMAP_IRQ dependency
  ...

9 years agoMerge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/nab/target...
Linus Torvalds [Sat, 20 Dec 2014 02:02:22 +0000 (18:02 -0800)]
Merge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/nab/target-pending

Pull SCSI target fixes from Nicholas Bellinger:
 "The highlights this merge window include:

   - Allow target fabric drivers to function as built-in.  (Roland)
   - Fix tcm_loop multi-TPG endpoint nexus bug.  (Hannes)
   - Move per device config_item_type into se_subsystem_api, allowing
     configfs attributes to be defined at module_init time.  (Jerome +
     nab)
   - Convert existing IBLOCK/FILEIO/RAMDISK/PSCSI/TCMU drivers to use
     external configfs attributes.  (nab)
   - A number of iser-target fixes related to active session + network
     portal shutdown stability during extended stress testing.  (Sagi +
     Slava)
   - Dynamic allocation of T10-PI contexts for iser-target, fixing a
     potentially bogus iscsi_np->tpg_np pointer reference in >= v3.14
     code.  (Sagi)
   - iser-target performance + scalability improvements.  (Sagi)
   - Fixes for SPC-4 Persistent Reservation AllRegistrants spec
     compliance.  (Ilias + James + nab)
   - Avoid potential short kern_sendmsg() in iscsi-target for now until
     Al's conversion to use msghdr iteration is merged post -rc1.
     (Viro)

  Also, Sagi has requested a number of iser-target patches (9) that
  address stability issues he's encountered during extended stress
  testing be considered for v3.10.y + v3.14.y code.  Given the amount of
  LOC involved, it will certainly require extra backporting effort.

  Apologies in advance to Greg-KH & Co on this.  Sagi and I will be
  working post-merge to ensure they each get applied correctly"

* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/nab/target-pending: (53 commits)
  target: Allow AllRegistrants to re-RESERVE existing reservation
  uapi/linux/target_core_user.h: fix headers_install.sh badness
  iscsi-target: Fail connection on short sendmsg writes
  iscsi-target: nullify session in failed login sequence
  target: Avoid dropping AllRegistrants reservation during unregister
  target: Fix R_HOLDER bit usage for AllRegistrants
  iscsi-target: Drop left-over bogus iscsi_np->tpg_np
  iser-target: Fix wc->wr_id cast warning
  iser-target: Remove code duplication
  iser-target: Adjust log levels and prettify some prints
  iser-target: Use debug_level parameter to control logging level
  iser-target: Fix logout sequence
  iser-target: Don't wait for session commands from completion context
  iser-target: Reduce CQ lock contention by batch polling
  iser-target: Introduce isert_poll_budget
  iser-target: Remove an atomic operation from the IO path
  iser-target: Remove redundant call to isert_conn_terminate
  iser-target: Use single CQ for TX and RX
  iser-target: Centralize completion elements to a context
  iser-target: Cast wr_id with uintptr_t instead of unsinged long
  ...

9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rkuo/linux...
Linus Torvalds [Sat, 20 Dec 2014 01:57:51 +0000 (17:57 -0800)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rkuo/linux-hexagon-kernel

Pull arch/hexagon updates from Richard Kuo:
 "Build cleanup and a few misc fixes"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rkuo/linux-hexagon-kernel:
  Hexagon: fix signal delivery for debug traps
  Hexagon: set ARCH_DMA_MINALIGN
  Hexagon: fix alignment of init_task in RW_DATA_SECTION
  hexagon: Fix build failures in linux-next

9 years agoMerge tag 'cris-changes-for-3.19' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 20 Dec 2014 01:08:14 +0000 (17:08 -0800)]
Merge tag 'cris-changes-for-3.19' of git://git.kernel.org/pub/scm/linux/kernel/git/jesper/cris

Pull arch/chris updates from Jesper Nilsson:
 "Mostly cleanup and build fixes for CRISv32 allmodconfig

  God Jul och Gott Nytt år!"

* tag 'cris-changes-for-3.19' of git://git.kernel.org/pub/scm/linux/kernel/git/jesper/cris:
  CRISv32: Remove last remnants of ETRAX_SPI_MMC_BOARD
  CRISv32: ETRAXFS: Fix recursive spinlock
  CRISv32: Select MTDRAM for axisflashmap
  CRISv32: Implement early console
  CRIS: Use KALLSYMs if available in call stack dump
  CRISv32: Fix declaration mismatch
  CRISv32: Rewrite of synchronous serial port driver
  CRIS: Update init memory handling
  CRISv32: Better handling of watchdog bite
  CRIS: Export missing function symbols
  CRIS: Export ioremap_nocache
  CRIS: Fix headers_install
  CRISv32: Add missing include for mm.h
  CRISv32: Drop obsolete file for SPI driver

9 years agoMerge tag 'please-pull-misc-3.19' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 20 Dec 2014 01:07:27 +0000 (17:07 -0800)]
Merge tag 'please-pull-misc-3.19' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux

Pull ia64 __get_cpu_var removal from Tony Luck:
 "__get_cpu_var removed from rest of tree, drop reference from comments
  in arch/ia64"

* tag 'please-pull-misc-3.19' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux:
  [IA64] Update comment that references __get_cpu_var

9 years agoCRISv32: Remove last remnants of ETRAX_SPI_MMC_BOARD
Jesper Nilsson [Wed, 29 Oct 2014 09:15:35 +0000 (10:15 +0100)]
CRISv32: Remove last remnants of ETRAX_SPI_MMC_BOARD

There are no users of this symbol left.

Reported-by: Paul Bolle <pebolle@tiscali.nl>
Signed-off-by: Jesper Nilsson <jesper.nilsson@axis.com>
9 years agoCRISv32: ETRAXFS: Fix recursive spinlock
Jesper Nilsson [Mon, 27 Oct 2014 08:11:31 +0000 (09:11 +0100)]
CRISv32: ETRAXFS: Fix recursive spinlock

Move pinmux alloc/dealloc code into functions that don't take
the spinlock so we can use from code that has the spinlock already.

CRISv32 has no working SMP, so spinlocks becomes a NOP,
so deadlock was never seen.

Signed-off-by: Jesper Nilsson <jesper.nilsson@axis.com>
9 years agoCRISv32: Select MTDRAM for axisflashmap
Jesper Nilsson [Mon, 27 Oct 2014 08:09:14 +0000 (09:09 +0100)]
CRISv32: Select MTDRAM for axisflashmap

Fixes compile error on allmodconfig.

Signed-off-by: Jesper Nilsson <jesper.nilsson@axis.com>
9 years agoCRISv32: Implement early console
Jesper Nilsson [Wed, 8 Oct 2014 13:52:49 +0000 (15:52 +0200)]
CRISv32: Implement early console

Signed-off-by: Jesper Nilsson <jesper.nilsson@axis.com>
9 years agoCRIS: Use KALLSYMs if available in call stack dump
Jesper Nilsson [Wed, 8 Oct 2014 12:57:35 +0000 (14:57 +0200)]
CRIS: Use KALLSYMs if available in call stack dump

Also, print kernel version on oops.

Signed-off-by: Jesper Nilsson <jesper.nilsson@axis.com>
9 years agoCRISv32: Fix declaration mismatch
Jesper Nilsson [Tue, 7 Oct 2014 11:42:35 +0000 (13:42 +0200)]
CRISv32: Fix declaration mismatch

Drop i2c_init from this header, it was declared non-static here,
but static in the C-file.

Signed-off-by: Jesper Nilsson <jesper.nilsson@axis.com>
9 years agoCRISv32: Rewrite of synchronous serial port driver
Jesper Nilsson [Tue, 7 Oct 2014 11:40:08 +0000 (13:40 +0200)]
CRISv32: Rewrite of synchronous serial port driver

Make driver possible to load as a module and try to handle
locking better.

Signed-off-by: Jesper Nilsson <jesper.nilsson@axis.com>
9 years agoCRIS: Update init memory handling
Jesper Nilsson [Tue, 7 Oct 2014 11:03:17 +0000 (13:03 +0200)]
CRIS: Update init memory handling

- Add free_initrd_mem as found by Guenter Roeck <linux@roeck-us.net>
- Add free_init_pages
- Export empty_zero_page symbol

Signed-off-by: Jesper Nilsson <jesper.nilsson@axis.com>
9 years agoCRISv32: Better handling of watchdog bite
Jesper Nilsson [Tue, 7 Oct 2014 10:59:24 +0000 (12:59 +0200)]
CRISv32: Better handling of watchdog bite

Don't enter watchdog handling if we're already in watchdog handling.

Also some minor formatting tweaks.

Signed-off-by: Jesper Nilsson <jesper.nilsson@axis.com>
9 years agoCRIS: Export missing function symbols
Jesper Nilsson [Tue, 7 Oct 2014 10:20:47 +0000 (12:20 +0200)]
CRIS: Export missing function symbols

strcmp was lost when all other string functions were removed,
but we still have an optimized version for this on CRISv32,
so any driver built as a module would not have access to this symbol.

In a similar manner, we had optimized versions of
csum_partial_copy_from_user and __do_clear_user
but no exported symbols for them, breaking bunch of other drivers
when built as a module.

At the same time, move EXPORT_SYMBOL(__copy_user) and
EXPORT_SYMBOL(__copy_user_zeroing) C-files so it's
located together with the function definition.

Signed-off-by: Jesper Nilsson <jesper.nilsson@axis.com>
9 years agoCRIS: Export ioremap_nocache
Jesper Nilsson [Wed, 1 Oct 2014 13:38:38 +0000 (15:38 +0200)]
CRIS: Export ioremap_nocache

Signed-off-by: Jesper Nilsson <jesper.nilsson@axis.com>
9 years agoCRIS: Fix headers_install
Sam Ravnborg [Mon, 14 Jul 2014 15:08:29 +0000 (17:08 +0200)]
CRIS: Fix headers_install

Fix headers_install by adjusting the path to arch files.
And delete unused Kbuild file.
Drop special handling of cris in the headers.sh script
as a nice side-effect.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Mikael Starvik <starvik@axis.com>
Cc: David Howells <dhowells@redhat.com>
Signed-off-by: Jesper Nilsson <jesper.nilsson@axis.com>
9 years agoCRISv32: Add missing include for mm.h
Jesper Nilsson [Tue, 30 Sep 2014 13:34:10 +0000 (15:34 +0200)]
CRISv32: Add missing include for mm.h

Fixes the following compile error.

arch/cris/arch-v32/kernel/time.c: In function 'reset_watchdog':
arch/cris/arch-v32/kernel/time.c:121:2:
        error: implicit declaration of function 'global_page_state'

Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Jesper Nilsson <jesper.nilsson@axis.com>
9 years agoCRISv32: Drop obsolete file for SPI driver
Jesper Nilsson [Tue, 30 Sep 2014 13:31:16 +0000 (15:31 +0200)]
CRISv32: Drop obsolete file for SPI driver

File was already deleted.

Signed-off-by: Jesper Nilsson <jesper.nilsson@axis.com>
9 years agoMerge branch 'x86-apic-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 19 Dec 2014 22:02:02 +0000 (14:02 -0800)]
Merge branch 'x86-apic-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull x86 apic updates from Thomas Gleixner:
 "After stopping the full x86/apic branch, I took some time to go
  through the first block of patches again, which are mostly cleanups
  and preparatory work for the irqdomain conversion and ioapic hotplug
  support.

  Unfortunaly one of the real problematic commits was right at the
  beginning, so I rebased this portion of the pending patches without
  the offenders.

  It would be great to get this into 3.19.  That makes reworking the
  problematic parts simpler.  The usual tip testing did not unearth any
  issues and it is fully bisectible now.

  I'm pretty confident that this wont affect the calmness of the xmas
  season.

  Changes:
   - Split the convoluted io_apic.c code into domain specific parts
     (vector, ioapic, msi, htirq)
   - Introduce proper helper functions to retrieve irq specific data
     instead of open coded dereferencing of pointers
   - Preparatory work for ioapic hotplug and irqdomain conversion
   - Removal of the non functional pci-ioapic driver
   - Removal of unused irq entry stubs
   - Make native_smp_prepare_cpus() preemtible to avoid GFP_ATOMIC
     allocations for everything which is called from there.
   - Small cleanups and fixes"

* 'x86-apic-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (36 commits)
  iommu/amd: Use helpers to access irq_cfg data structure associated with IRQ
  iommu/vt-d: Use helpers to access irq_cfg data structure associated with IRQ
  x86: irq_remapping: Use helpers to access irq_cfg data structure associated with IRQ
  x86, irq: Use helpers to access irq_cfg data structure associated with IRQ
  x86, irq: Make MSI and HT_IRQ indepenent of X86_IO_APIC
  x86, irq: Move IRQ initialization routines from io_apic.c into vector.c
  x86, irq: Move IOAPIC related declarations from hw_irq.h into io_apic.h
  x86, irq: Move HT IRQ related code from io_apic.c into htirq.c
  x86, irq: Move PCI MSI related code from io_apic.c into msi.c
  x86, irq: Replace printk(KERN_LVL) with pr_lvl() utilities
  x86, irq: Make UP version of irq_complete_move() an inline stub
  x86, irq: Move local APIC related code from io_apic.c into vector.c
  x86, irq: Introduce helpers to access struct irq_cfg
  x86, irq: Protect __clear_irq_vector() with vector_lock
  x86, irq: Rename local APIC related functions in io_apic.c as apic_xxx()
  x86, irq: Refine hw_irq.h to prepare for irqdomain support
  x86, irq: Convert irq_2_pin list to generic list
  x86, irq: Kill useless parameter 'irq_attr' of IO_APIC_get_PCI_irq_vector()
  x86, irq, acpi: Get rid of special handling of GSI for ACPI SCI
  x86, irq: Introduce helper to check whether an IOAPIC has been registered
  ...

9 years agoPM: Eliminate CONFIG_PM_RUNTIME
Rafael J. Wysocki [Fri, 19 Dec 2014 14:37:54 +0000 (15:37 +0100)]
PM: Eliminate CONFIG_PM_RUNTIME

Having switched over all of the users of CONFIG_PM_RUNTIME to use
CONFIG_PM directly, turn the latter into a user-selectable option
and drop the former entirely from the tree.

Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Acked-by: Kevin Hilman <khilman@linaro.org>