linux-stable.git/drivers/crypto/caam, branch linux-4.3.y

crypto: caam - make write transactions bufferable on PPC platforms

2016-02-19T22:28:36+00:00

commit e7a7104e432c0db8469ca3568daf4f1d1afe3e73 upstream.

Previous change (see "Fixes" tag) to the MCFGR register
clears AWCACHE[0] ("bufferable" AXI3 attribute) (which is "1" at POR).

This makes all writes non-bufferable, causing a ~ 5% performance drop
for PPC-based platforms.

Rework previous change such that MCFGR[AWCACHE]=4'b0011
(bufferable + cacheable) for all platforms.
Note: For ARM-based platforms, AWCACHE[0] is ignored
by the interconnect IP.

Fixes: f10967495144 ("crypto: caam - fix snooping for write transactions")
Signed-off-by: Horia Geant? 
Signed-off-by: Herbert Xu 
Signed-off-by: Greg Kroah-Hartman

crypto: caam - fix non-block aligned hash calculation

2016-02-19T22:28:22+00:00

commit c7556ff7e3e4f2747583bcc787f12ec9460ec3a6 upstream.

caam does not properly calculate the size of the retained state
when non-block aligned hashes are requested - it uses the wrong
buffer sizes, which results in errors such as:

caam_jr 2102000.jr1: 40000501: DECO: desc idx 5: SGT Length Error. The descriptor is trying to read more data than is contained in the SGT table.

We end up here with:

in_len 0x46 blocksize 0x40 last_bufsize 0x0 next_bufsize 0x6
to_hash 0x40 ctx_len 0x28 nbytes 0x20

which results in a job descriptor of:

jobdesc@889: ed03d918: b0861c08 3daa0080 f1400000 3d03d938
jobdesc@889: ed03d928: 00000068 f8400000 3cde2a40 00000028

where the word at 0xed03d928 is the expected data size (0x68), and a
scatterlist containing:

sg@892: ed03d938: 00000000 3cde2a40 00000028 00000000
sg@892: ed03d948: 00000000 3d03d100 00000006 00000000
sg@892: ed03d958: 00000000 7e8aa700 40000020 00000000

0x68 comes from 0x28 (the context size) plus the "in_len" rounded down
to a block size (0x40).  in_len comes from 0x26 bytes of unhashed data
from the previous operation, plus the 0x20 bytes from the latest
operation.

The fixed version would create:

sg@892: ed03d938: 00000000 3cde2a40 00000028 00000000
sg@892: ed03d948: 00000000 3d03d100 00000026 00000000
sg@892: ed03d958: 00000000 7e8aa700 40000020 00000000

which replaces the 0x06 length with the correct 0x26 bytes of previously
unhashed data.

This fixes a previous commit which erroneously "fixed" this due to a
DMA-API bug report; that commit indicates that the bug was caused via a
test_ahash_pnum() function in the tcrypt module.  No such function has
ever existed in the mainline kernel.  Given that the change in this
commit has been tested with DMA API debug enabled and shows no issue,
I can only conclude that test_ahash_pnum() was triggering that bad
behaviour by CAAM.

Fixes: 7d5196aba3c8 ("crypto: caam - Correct DMA unmap size in ahash_update_ctx()")
Signed-off-by: Russell King 
Signed-off-by: Herbert Xu 
Signed-off-by: Greg Kroah-Hartman

Merge git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6

2015-09-01T00:38:39+00:00

Pull crypto updates from Herbert Xu:
 "Here is the crypto update for 4.3:

  API:

   - the AEAD interface transition is now complete.
   - add top-level skcipher interface.

  Drivers:

   - x86-64 acceleration for chacha20/poly1305.
   - add sunxi-ss Allwinner Security System crypto accelerator.
   - add RSA algorithm to qat driver.
   - add SRIOV support to qat driver.
   - add LS1021A support to caam.
   - add i.MX6 support to caam"

* git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6: (163 commits)
  crypto: algif_aead - fix for multiple operations on AF_ALG sockets
  crypto: qat - enable legacy VFs
  MPI: Fix mpi_read_buffer
  crypto: qat - silence a static checker warning
  crypto: vmx - Fixing opcode issue
  crypto: caam - Use the preferred style for memory allocations
  crypto: caam - Propagate the real error code in caam_probe
  crypto: caam - Fix the error handling in caam_probe
  crypto: caam - fix writing to JQCR_MS when using service interface
  crypto: hash - Add AHASH_REQUEST_ON_STACK
  crypto: testmgr - Use new skcipher interface
  crypto: skcipher - Add top-level skcipher interface
  crypto: cmac - allow usage in FIPS mode
  crypto: sahara - Use dmam_alloc_coherent
  crypto: caam - Add support for LS1021A
  crypto: qat - Don't move data inside output buffer
  crypto: vmx - Fixing GHASH Key issue on little endian
  crypto: vmx - Fixing AES-CTR counter bug
  crypto: null - Add missing Kconfig tristate for NULL2
  crypto: nx - Add forward declaration for struct crypto_aead
  ...

crypto: caam - Use the preferred style for memory allocations

2015-08-24T14:07:42+00:00

"The preferred form for passing a size of a struct is the following:

        p = kmalloc(sizeof(*p), ...);
....

The preferred form for allocating a zeroed array is the following:

        p = kcalloc(n, sizeof(...), ...); "

,so do as suggested.

Signed-off-by: Fabio Estevam 
Reviewed-by: Horia Geant? 
Signed-off-by: Herbert Xu

crypto: caam - Propagate the real error code in caam_probe

2015-08-24T14:07:41+00:00

Instead of propagating a 'fake' error code, just propagate the real
one in the case of caam_drv_identify_clk() failure.

Signed-off-by: Fabio Estevam 
Reviewed-by: Horia Geant? 
Signed-off-by: Herbert Xu

crypto: caam - Fix the error handling in caam_probe

2015-08-24T14:07:39+00:00

In the error path we should disable the resources that were previously
acquired, so fix the error handling accordingly.

Signed-off-by: Fabio Estevam 
Reviewed-by: Horia Geant? 
Signed-off-by: Herbert Xu

crypto: caam - fix writing to JQCR_MS when using service interface

2015-08-24T14:07:38+00:00

Most significant part of JQCR (Job Queue Control Register) contains
bits that control endianness: ILE - Immediate Little Endian,
DWS - Double Word Swap.
The bits are automatically set by the Job Queue Controller HW.

Unfortunately these bits are cleared in SW when submitting descriptors
via the register-based service interface.
>From LS1021A:
JQCR_MS = 08080100 - before writing: ILE | DWS | SRC (JR0)
JQCR_MS = 30000100 - after writing: WHL | FOUR | SRC (JR0)

This would cause problems on little endian caam for descriptors
containing immediata data or double-word pointers.
Currently there is no problem since the only descriptors ran through
this interface are the ones that (un)instantiate RNG.

Signed-off-by: Horia Geant? 
Signed-off-by: Herbert Xu

crypto: caam - add support for LS1021A

2015-08-18T02:30:39+00:00

LS1021A is a QorIQ SoC having little endian CAAM.

There are a few differences b/w QorIQ and i.MX from CAAM perspective:

1. i.MX platforms are somewhat special wrt. 64-bit registers:
-big endian format at 64-bit level: MSW at address+0 and LSW at address+4
-little endian format at 32-bit level (within MSW and LSW)
and thus need special handling.

2. No CCM (clock controller module) for QorIQ.
No CAAM clocks to enable / disable.

A new Kconfig option - CRYPTO_DEV_FSL_CAAM_LE - is added to indicate
CAAM is little endian (*). It is hidden from the user (to avoid
misconfiguration); when adding support for a new platform with LE CAAM,
either the Kconfig needs to be updated or the corresponding defconfig
needs to indicate that CAAM is LE.
(*) Using a DT property to provide CAAM endianness would not allow
for the ifdeffery.

In order to keep changes to a minimum, the following changes
are postponed:
-endianness fix of the last word in the S/G (rsvd2, bpid, offset),
fields are always 0 anyway;
-S/G format fix for i.MX7 (yes, i.MX7 support was not added yet,
but still...)

Signed-off-by: Horia Geant? 
Signed-off-by: Herbert Xu

crypto: aead - Remove CRYPTO_ALG_AEAD_NEW flag

2015-08-17T08:53:53+00:00

This patch removes the CRYPTO_ALG_AEAD_NEW flag now that everyone
has been converted.

Signed-off-by: Herbert Xu

crypto: caam - Remove unneeded 'ret' variable

2015-08-13T07:13:29+00:00

Variable 'ret' is only used for returning the value 0.

We can make it simpler and just return 0 instead.

The semantic patch that makes this change is available
in scripts/coccinelle/misc/returnvar.cocci.

Signed-off-by: Fabio Estevam 
Signed-off-by: Herbert Xu