Wolfgang Walter reported this oops on his via C3 using padlock for
AES-encryption:
##################################################################
BUG: unable to handle kernel NULL pointer dereference at 000001f0
IP: [<c01028c5>] __switch_to+0x30/0x117
*pde = 00000000
Oops: 0002 [#1] PREEMPT
Modules linked in:
Pid: 2071, comm: sleep Not tainted (2.6.26 #11)
EIP: 0060:[<c01028c5>] EFLAGS: 00010002 CPU: 0
EIP is at __switch_to+0x30/0x117
EAX: 00000000 EBX: c0493300 ECX: dc48dd00 EDX: c0493300
ESI: dc48dd00 EDI: c0493530 EBP: c04cff8c ESP: c04cff7c
DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 0068
Process sleep (pid: 2071, ti=c04ce000 task=dc48dd00 task.ti=d2fe6000)
Stack: dc48df30 c0493300 00000000 00000000 d2fe7f44 c03b5b43 c04cffc8 00000046
c0131856 0000005a dc472d3c c0493300 c0493470 d983ae00 00002696 00000000
c0239f54 00000000 c04c4000 c04cffd8 c01025fe c04f3740 00049800 c04cffe0
Call Trace:
[<c03b5b43>] ? schedule+0x285/0x2ff
[<c0131856>] ? pm_qos_requirement+0x3c/0x53
[<c0239f54>] ? acpi_processor_idle+0x0/0x434
[<c01025fe>] ? cpu_idle+0x73/0x7f
[<c03a4dcd>] ? rest_init+0x61/0x63
=======================
Wolfgang also found out that adding kernel_fpu_begin() and kernel_fpu_end()
around the padlock instructions fix the oops.
Suresh wrote:
These padlock instructions though don't use/touch SSE registers, but it behaves
similar to other SSE instructions. For example, it might cause DNA faults
when cr0.ts is set. While this is a spurious DNA trap, it might cause
oops with the recent fpu code changes.
This is the code sequence that is probably causing this problem:
a) new app is getting exec'd and it is somewhere in between
start_thread() and flush_old_exec() in the load_xyz_binary()
b) At pont "a", task's fpu state (like TS_USEDFPU, used_math() etc) is
cleared.
c) Now we get an interrupt/softirq which starts using these encrypt/decrypt
routines in the network stack. This generates a math fault (as
cr0.ts is '1') which sets TS_USEDFPU and restores the math that is
in the task's xstate.
d) Return to exec code path, which does start_thread() which does
free_thread_xstate() and sets xstate pointer to NULL while
the TS_USEDFPU is still set.
e) At the next context switch from the new exec'd task to another task,
we have a scenarios where TS_USEDFPU is set but xstate pointer is null.
This can cause an oops during unlazy_fpu() in __switch_to()
Now:
1) This should happen with or with out pre-emption. Viro also encountered
similar problem with out CONFIG_PREEMPT.
2) kernel_fpu_begin() and kernel_fpu_end() will fix this problem, because
kernel_fpu_begin() will manually do a clts() and won't run in to the
situation of setting TS_USEDFPU in step "c" above.
3) This was working before the fpu changes, because its a spurious
math fault which doesn't corrupt any fpu/sse registers and the task's
math state was always in an allocated state.
With out the recent lazy fpu allocation changes, while we don't see oops,
there is a possible race still present in older kernels(for example,
while kernel is using kernel_fpu_begin() in some optimized clear/copy
page and an interrupt/softirq happens which uses these padlock
instructions generating DNA fault).
This is the failing scenario that existed even before the lazy fpu allocation
changes:
0. CPU's TS flag is set
1. kernel using FPU in some optimized copy routine and while doing
kernel_fpu_begin() takes an interrupt just before doing clts()
2. Takes an interrupt and ipsec uses padlock instruction. And we
take a DNA fault as TS flag is still set.
3. We handle the DNA fault and set TS_USEDFPU and clear cr0.ts
4. We complete the padlock routine
5. Go back to step-1, which resumes clts() in kernel_fpu_begin(), finishes
the optimized copy routine and does kernel_fpu_end(). At this point,
we have cr0.ts again set to '1' but the task's TS_USEFPU is stilll
set and not cleared.
6. Now kernel resumes its user operation. And at the next context
switch, kernel sees it has do a FP save as TS_USEDFPU is still set
and then will do a unlazy_fpu() in __switch_to(). unlazy_fpu()
will take a DNA fault, as cr0.ts is '1' and now, because we are
in __switch_to(), math_state_restore() will get confused and will
restore the next task's FP state and will save it in prev tasks's FP state.
Remember, in __switch_to() we are already on the stack of the next task
but take a DNA fault for the prev task.
This causes the fpu leakage.
Fix the padlock instruction usage by calling them inside the
context of new routines irq_ts_save/restore(), which clear/restore cr0.ts
manually in the interrupt context. This will not generate spurious DNA
in the context of the interrupt which will fix the oops encountered and
the possible FPU leakage issue.
Reported-and-bisected-by: Wolfgang Walter <wolfgang.walter@stwm.de>
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Later SEC revision requires the link table (used for scatter/gather)
to have an extra entry to account for the total length in descriptor [4],
which contains cipher Input and ICV.
This only applies to decrypt, not encrypt.
Without this change, on 837x, a gather return/length error results
when a decryption uses a link table to gather the fragments.
This is observed by doing a ping with size of 1447 or larger with AES,
or a ping with size 1455 or larger with 3des.
So, add check for SEC compatible "fsl,3.0" for using extra link table entry.
Signed-off-by: Lee Nipper <lee.nipper@freescale.com>
Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
free edescriptor when returning error (such as -EAGAIN).
Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
Acked-by: Lee Nipper <lee.nipper@freescale.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
use GFP_ATOMIC when necessary; use atomic_t when allocating submit_count.
Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
Acked-by: Lee Nipper <lee.nipper@freescale.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
add requests pending/submit count to prevent request queue full
condition by preempting h/w overflow interrupts in software.
We do this due to the delay in the delivery and handling of the
channel overflow error interrupt.
Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
Acked-by: Lee Nipper <lee.nipper@freescale.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Seems that dst == src, but this fixes the logic in case it's not.
Signed-off-by: Lee Nipper <lee.nipper@freescale.com>
Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Remove of_node_put calls since there is no corresponding of_node_get.
This patch prevents an exception when talitos is loaded a 2nd time.
This sequence: modprobe talitos; rmmod talitos; modprobe talitos
causes this message: "WARNING: Bad of_node_put() on /soc8349@e0000000/crypto@30000".
Signed-off-by: Lee Nipper <lee.nipper@freescale.com>
Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Without CRYPTO_AUTHENC the driver fails to build:
drivers/built-in.o: In function `ixp_module_init':
ixp4xx_crypto.c:(.init.text+0x3250): undefined reference to `crypto_aead_type'
Signed-off-by: Imre Kaloz <kaloz@openwrt.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
This patch adds support for:
authenc(hmac(sha256),cbc(aes)),
authenc(hmac(sha256),cbc(des3_ede)),
authenc(hmac(md5),cbc(aes)),
authenc(hmac(md5),cbc(des3_ede)).
Some constant usage was changed to use aes, des, and sha include files.
Signed-off-by: Lee Nipper <lee.nipper@freescale.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
The name authenc(hmac(sha1-talitos),cbc(aes-talitos)) is potentially
ambiguous since it could also mean using the generic authenc template
on hmac(sha1-talitos) and cbc(aes-talitos). In general, parentheses
should be reserved for templates that spawn algorithms.
This patches changes it to the form authenc-hmac-sha1-cbc-aes-talitos.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
This patch adds support for authenc(hmac(sha1),cbc(des3_ede))
to the talitos crypto driver for the Freescale Security Engine.
Some adjustments were made to the scatterlist to link table conversion
to make 3des work for ping -s 1439..1446.
Signed-off-by: Lee Nipper <lee.nipper@freescale.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
When loading aes or sha256 via the module aliases, the padlock modules
also try to get loaded. Make the error message for them not being
present only be a NOTICE rather than an ERROR so that use of 'quiet'
will suppress the messages
Signed-off-by: Jeremy Katz <katzj@redhat.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Add support for the hardware crypto engine provided by the NPE C
of the Intel IXP4xx networking processor series.
Supported ciphers: des, des3, aes
and a combination of them with md5 and sha1 hmac
Signed-off-by: Christian Hohnstaedt <chohnstaedt@innominate.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Add support for the SEC available on a wide range of PowerQUICC devices,
e.g. MPC8349E, MPC8548E.
This initial version supports authenc(hmac(sha1),cbc(aes)) for use with IPsec.
Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
The descriptors need to be invalidated after processing for ring
cleanup to work properly and to avoid using an old destination
descriptor when the src and cmd descriptors are already set up
and the dst descriptor isn't.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Acked-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Move command descriptor setup to seperate function as preparation
for the following DMA setup fixes.
Note 1: also fix a harmless typo while moving it: sa_idx is initialized
to dma->resi instead of dma->cmdi.
Note 2: errors from command descriptor setup are not propagated back,
anymore, they can't be handled anyway and all conditions leading
to errors should be checked earlier.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Acked-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
All but the last element of the command and result descriptor rings can be
used for crypto requests, fix HIFN_QUEUE_LENGTH.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Acked-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
For combined modes like cbc(aes) the driver is responsible for
initializing ivsize.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
When the CryptoAPI instantiates a new algorithm, it performs a lookup
by driver name. Since hifn uses the same name for all modes of one
algorithm, the lookup may return an incorrect algorithm.
Change the name to use <mode>-<algo>-<devicename> to provide unique
names for the different combinations and devices.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Acked-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
The scatterlist may contain more data than the crypto request, causing
an underflow of the remaining byte count while walking the list.
Use the minimum of the scatterlist element size and the remaining byte
count specified in the crypto request to avoid this.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Acked-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
The check for misalignment of the scatterlist data has two bugs:
- the source buffer doesn't need to be aligned at all
- the destination buffer and its size needs to be aligned to a multiple
of 4, not to the crypto alg blocksize
Introduce symbolic constant for destination buffer alignment requirements,
use it instead of the crypto alg blocksize and remove the unnecessary
checks for source buffer alignment and change cra_alignmask to zero.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
ablkcipher_walk may return a negative error value, handle this properly
instead of treating it as a huge number of scatter-gather elements.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Acked-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
hifn_setup_crypto() needs to return -EINPROGRESS on success to indicate
asynchronous processing to the crypto API. This also means it must not
return the errno code returned by hifn_process_queue(), if any.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Without debugging this spams the log with "printk: N messages surpressed"
without any actual messages on error. With debugging its more useful to
always see the message.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Acked-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
HIFN uses little-endian by default, move cpu_to_le32 conversion to hifn_write_0/
hifn_write_1, add sparse annotations and fix an invalid endian conversion in
hifn_setup_src_desc.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Acked-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
The Padlock AES setkey routine is the same as exported by the generic
implementation. So we could use it.
Signed-off-by: Sebastian Siewior <sebastian@breakpoint.cc>
Cc: Michal Ludvig <michal@logix.cz>
Tested-by: Stefan Hellermann <stefan@the2masters.de>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Exploit the System z10 hardware acceleration for SHA384.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Exploit the System z10 hardware acceleration for SHA512.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
This patch allows user space applications to access large amounts of
truly random data. The random data source is the build-in hardware
random number generator on the CEX2C cards.
Signed-off-by: Ralph Wuerthner <rwuerthn@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
NB: remaining endianness warnings in the file are, AFAICS, real bugs.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The CRYPTO_DEV_HIFN_795X_RNG ifdefs are missing the CONFIG_ prefix.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Move s390 crypto Kconfig options to drivers/crypto/Kconfig to have all
hardware crypto devices in one place.
This also makes messing up the kernel source tree easier for some people.
Signed-off-by: Jan Glauber <jan.glauber@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Currently it is possible to select HW_RANDOM as a module and have
hifn795x built-in. This causes a build problem because hifn795x
will then call hwrng_register which isn't built-in.
This patch introduces a new config option to control the hifn795x
RNG which lets us avoid this problem.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
* git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6: (125 commits)
[CRYPTO] twofish: Merge common glue code
[CRYPTO] hifn_795x: Fixup container_of() usage
[CRYPTO] cast6: inline bloat--
[CRYPTO] api: Set default CRYPTO_MINALIGN to unsigned long long
[CRYPTO] tcrypt: Make xcbc available as a standalone test
[CRYPTO] xcbc: Remove bogus hash/cipher test
[CRYPTO] xcbc: Fix algorithm leak when block size check fails
[CRYPTO] tcrypt: Zero axbuf in the right function
[CRYPTO] padlock: Only reset the key once for each CBC and ECB operation
[CRYPTO] api: Include sched.h for cond_resched in scatterwalk.h
[CRYPTO] salsa20-asm: Remove unnecessary dependency on CRYPTO_SALSA20
[CRYPTO] tcrypt: Add select of AEAD
[CRYPTO] salsa20: Add x86-64 assembly version
[CRYPTO] salsa20_i586: Salsa20 stream cipher algorithm (i586 version)
[CRYPTO] gcm: Introduce rfc4106
[CRYPTO] api: Show async type
[CRYPTO] chainiv: Avoid lock spinning where possible
[CRYPTO] seqiv: Add select AEAD in Kconfig
[CRYPTO] scatterwalk: Handle zero nbytes in scatterwalk_map_and_copy
[CRYPTO] null: Allow setkey on digest_null
...
Currently we reset the key for each segment fed to the xcrypt instructions.
This patch optimises this for CBC and ECB so that we only do this once for
each encrypt/decrypt operation.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
With the impending addition of the givcipher type, both blkcipher and
ablkcipher algorithms will use it to create givcipher objects. As such
it no longer makes sense to split the system between ablkcipher and
blkcipher. In particular, both ablkcipher.c and blkcipher.c would need
to use the givcipher type which has to reside in ablkcipher.c since it
shares much code with it.
This patch merges the two Kconfig options as well as the modules into one.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
crypto_blkcipher_decrypt is wrong because it does not care about
the IV.
Signed-off-by: Sebastian Siewior <sebastian@breakpoint.cc>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
There is no reason to keep the IV in the private structre. Instead keep
just a pointer to make the patch smaller :)
This also remove a few memcpy()s
Signed-off-by: Sebastian Siewior <sebastian@breakpoint.cc>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
The current PLL initalization has a number of deficiencies:
- uses fixed multiplier of 8, which overclocks the chip when using a
reference clock that operates at frequencies above 33MHz. According
to a comment in the BSD source, this is true for the external clock
on almost all every board.
- writes to a reserved bit
- doesn't follow the initialization procedure specified in chapter
6.11.1 of the HIFN hardware users guide
- doesn't allow to use the PCI clock
This patch adds a module parameter to specify the reference clock
(pci or external) and its frequency and uses that to calculate the
optimum multiplier to reach the maximal speed. By default it uses
the external clock and assumes a speed of 66MHz, which effectively
halfs the frequency currently used.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Acked-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Up until now we have ablkcipher algorithms have been identified as
type BLKCIPHER with the ASYNC bit set. This is suboptimal because
ablkcipher refers to two things. On the one hand it refers to the
top-level ablkcipher interface with requests. On the other hand it
refers to and algorithm type underneath.
As it is you cannot request a synchronous block cipher algorithm
with the ablkcipher interface on top. This is a problem because
we want to be able to eventually phase out the blkcipher top-level
interface.
This patch fixes this by making ABLKCIPHER its own type, just as
we have distinct types for HASH and DIGEST. The type it associated
with the algorithm implementation only.
Which top-level interface is used for synchronous block ciphers is
then determined by the mask that's used. If it's a specific mask
then the old blkcipher interface is given, otherwise we go with the
new ablkcipher interface.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
alpha:
drivers/crypto/hifn_795x.c: In function 'ablkcipher_walk_init':
drivers/crypto/hifn_795x.c:1231: error: implicit declaration of function 'sg_init_table'
drivers/crypto/hifn_795x.c:1243: error: implicit declaration of function 'sg_set_page'
drivers/crypto/hifn_795x.c: In function 'ablkcipher_walk_exit':
drivers/crypto/hifn_795x.c:1257: error: implicit declaration of function 'sg_page'
drivers/crypto/hifn_795x.c:1257: warning: passing argument 1 of '__free_pages' makes pointer from integer without a cast
drivers/crypto/hifn_795x.c: In function 'ablkcipher_add':
drivers/crypto/hifn_795x.c:1278: warning: passing argument 1 of 'kmap_atomic' makes pointer from integer without a cast
drivers/crypto/hifn_795x.c: In function 'ablkcipher_walk':
drivers/crypto/hifn_795x.c:1336: warning: passing argument 1 of 'kmap_atomic' makes pointer from integer without a cast
drivers/crypto/hifn_795x.c: In function 'hifn_setup_session':
drivers/crypto/hifn_795x.c:1465: warning: assignment makes pointer from integer without a cast
drivers/crypto/hifn_795x.c:1469: warning: assignment makes pointer from integer without a cast
drivers/crypto/hifn_795x.c:1472: warning: assignment makes pointer from integer without a cast
drivers/crypto/hifn_795x.c: In function 'ablkcipher_get':
drivers/crypto/hifn_795x.c:1593: warning: passing argument 1 of 'kmap_atomic' makes pointer from integer without a cast
{standard input}: Assembler messages:
{standard input}:7: Warning: setting incorrect section attributes for .got
drivers/crypto/hifn_795x.c: In function 'hifn_process_ready':
drivers/crypto/hifn_795x.c:1653: warning: passing argument 1 of 'kmap_atomic' makes pointer from integer without a cast
drivers/crypto/hifn_795x.c: In function 'hifn_probe':
drivers/crypto/hifn_795x.c:2438: error: 'DMA_32BIT_MASK' undeclared (first use in this function)
drivers/crypto/hifn_795x.c:2438: error: (Each undeclared identifier is reported only once
drivers/crypto/hifn_795x.c:2438: error: for each function it appears in.)
drivers/crypto/hifn_795x.c:2443: warning: format '%d' expects type 'int', but argument 4 has type 'long int'
drivers/crypto/hifn_795x.c:2443: warning: format '%d' expects type 'int', but argument 4 has type 'long int'
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
The HIFN driver is currently selectable on s390 but wont compile.
Since it looks like HIFN needs PCI make the Kconfig dependent on PCI,
which is not available on s390.
Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Acked-by: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Suresh Siddha