Commit Graph

135 Commits

Author SHA1 Message Date
jmg
584181912d const'ify an arg that we don't update... 2015-07-29 23:37:15 +00:00
jmg
ec1fe94d10 Complete the move that was started w/ r263218.. For some reason I
didn't delete the files, so that means we need to bring the changes in
r282726 to the correct files..

make tinderbox completed with this patch...
2015-07-11 03:12:34 +00:00
jmg
d374814126 upon further examination, it turns out that _unregister_all already
provides the guarantee that no threads will be in the _newsession code..
This is provided by the CRYPTODRIVER lock...  This makes the pause
unneeded...
2015-07-08 22:48:41 +00:00
jmg
1ca19e1938 address an issue where consumers, like IPsec, can reuse the same
session in multiple threads w/o locking..  There was a single fpu
context shared per session, if multiple threads were using the session,
and both migrated away, they could corrupt each other's fpu context...

This patch adds a per cpu context and a lock to protect it...

It also tries to better address unloading of the aesni module...
The pause will be removed once the OpenCrypto Framework provides a
better method for draining callers into _newsession...

I first discovered the fpu context sharing issue w/ a flood ping over
an IPsec tunnel between two bhyve machines...  The patch in D3015
was used to verify that this fix does fix the issue...

Reviewed by:	gnn, kib (both earlier versions)
Differential Revision:        https://reviews.freebsd.org/D3016
2015-07-08 19:15:29 +00:00
jmg
2b00c9ed0d unroll the loop slightly... This improves performance enough to
justify, especially for CBC performance where we can't pipeline..  I
don't happen to have my measurements handy though...

Sponsored by:	Netflix, Inc.
2015-07-07 20:31:09 +00:00
jmg
abd8f81b92 Fix for non-random IV's when CRD_F_IV_PRESENT and CRD_F_IV_EXPLICIT
flags are not specified... This bug was introduced in r275732...

This only affects IPsec ESP only policies w/ the aesni module loaded,
other subsystems specify one or both of the flags...

Reviewed by:	gnn, delphij, eri
2015-07-06 19:30:29 +00:00
jmg
384ca830a8 remove _NORMAL flag which isn't suppose to be used w/ _alloc_ctx...
Reviewed by:	kib (a while ago)
2015-07-06 19:17:56 +00:00
rodrigc
5e984ef151 In the version of gcc in the FreeBSD tree, this modification was made to
the compiler in svn r242182:

#if STDC_HOSTED
#include <mm_malloc.h>
#endif

A similar change was done to clang in the FreeBSD tree in svn r218893:

However, for external gcc toolchains, this patch is not in the compiler's header
file.

This patch to FreeBSD's aesni code allows compilation with an external
gcc toolchain.

Differential Revision: https://reviews.freebsd.org/D2285
Reviewed by:  jmg, dim
Approved by:  dim
2015-04-16 17:42:52 +00:00
jmg
c3ff54cc39 Add some new modes to OpenCrypto. These modes are AES-ICM (can be used
for counter mode), and AES-GCM.  Both of these modes have been added to
the aesni module.

Included is a set of tests to validate that the software and aesni
module calculate the correct values.  These use the NIST KAT test
vectors.  To run the test, you will need to install a soon to be
committed port, nist-kat that will install the vectors.  Using a port
is necessary as the test vectors are around 25MB.

All the man pages were updated.  I have added a new man page, crypto.7,
which includes a description of how to use each mode.  All the new modes
and some other AES modes are present.  It would be good for someone
else to go through and document the other modes.

A new ioctl was added to support AEAD modes which AES-GCM is one of them.
Without this ioctl, it is not possible to test AEAD modes from userland.

Add a timing safe bcmp for use to compare MACs.  Previously we were using
bcmp which could leak timing info and result in the ability to forge
messages.

Add a minor optimization to the aesni module so that single segment
mbufs don't get copied and instead are updated in place.  The aesni
module needs to be updated to support blocked IO so segmented mbufs
don't have to be copied.

We require that the IV be specified for all calls for both GCM and ICM.
This is to ensure proper use of these functions.

Obtained from:	p4: //depot/projects/opencrypto
Relnotes:	yes
Sponsored by:	FreeBSD Foundation
Sponsored by:	NetGate
2014-12-12 19:56:36 +00:00
des
ce96d70e17 Fix gcc build: preserve const qualifier when casting input values.
Noticed by:	bz@
Submitted by:	dim@
2014-11-11 13:37:28 +00:00
des
d656003ded Constify the AES code and propagate to consumers. This allows us to
update the Fortuna code to use SHAd-256 as defined in FS&K.

Approved by:	so (self)
2014-11-10 09:44:38 +00:00
kib
4b355196ab Put the aesni_cipher_setup() and aesni_cipher_process() functions into
the file which is compiled with SSE disabled.  The functions set up
the FPU context for kernel, and compiler optimizations which could
lead to use of XMM registers before the fpu_kern_enter(9) is called or
after fpu_kern_leave(9), panic the machine.

Discussed with:	jmg
Sponsored by:	The FreeBSD Foundation
MFC after:	1 week
2014-06-24 06:55:49 +00:00
kib
fe547198b1 Add FPU_KERN_KTHR flag to fpu_kern_enter(9), which avoids saving FPU
context into memory for the kernel threads which called
fpu_kern_thread(9).  This allows the fpu_kern_enter() callers to not
check for is_fpu_kern_thread() to get the optimization.

Apply the flag to padlock(4) and aesni(4).  In aesni_cipher_process(),
do not leak FPU context state on error.

Sponsored by:	The FreeBSD Foundation
MFC after:	1 week
2014-06-23 07:37:54 +00:00
imp
c39e6fc2c9 NO_MAN= has been deprecated in favor of MAN= for some time, go ahead
and finish the job. ncurses is now the only Makefile in the tree that
uses it since it wasn't a simple mechanical change, and will be
addressed in a future commit.
2014-04-13 05:21:56 +00:00
jmg
eab153cc9d replace the kernel's version w/ cperciva's implementation... In all
my tests, it is faster ~20%, even on an old IXP425 533MHz it is ~45%
faster...  This is partly due to loop unrolling, so the code size does
significantly increase...  I do plan on committing a version that
rolls up the loops again for smaller code size for embedded systems
where size is more important than absolute performance (it'll save ~6k
code)...

The kernel implementation is now shared w/ userland's libcrypt and
libmd...

We drop support for sha256 from sha2.c, so now sha2.c only contains
sha384 and sha512...

Reviewed by:	secteam@
2014-03-16 01:43:23 +00:00
jmg
bf2e9f212d copy these files from lib/libmd in preperation for moving these files
into the kernel...
2014-03-16 00:57:26 +00:00
jmg
f19a97b48c fix broken style(9) in r258399
Pointed out by:	brd
2013-11-23 00:28:18 +00:00
jmg
472b0e6b78 flag that the aesni driver is sync... This means we don't waste a
context switch just to call the done callback...  On my machine, this
improves geli/gzero decrypt performance by ~27% from 550MB/sec to
~700MB/sec...

MFC after:	3 days
2013-11-20 20:25:27 +00:00
jmg
c4e415a9b1 make it so that from/to can be missaligned as it can happen (the geli
regression manages to do it)...  We use a packed struct to coerce
gcc/clang into producing unaligned loads (there is not packed pointer
attribute, otherwise this would be easier)...

use _storeu_ and _loadu_ when using the structure is overkill...

be better at using types properly...  Since we allocate our own key
schedule and make sure it's aligned, use the __m128i type in various
arguments to functions...

clang ignores __aligned on prototypes and gcc errors on them, leave them
in comments to document that these function arguments are require to be
aligned...

about all that changes is movdqa -> movdqu from reading the diff of the
disassembly output...

Noticed by:	symbolics at gmx.com
MFC after:	3 days
2013-11-06 19:14:49 +00:00
jmg
8ec798134a Use the fact that the AES-NI instructions can be pipelined to improve
performance... Use SSE2 instructions for calculating the XTS tweek
factor...  Let the compiler do more work and handle register allocation
by using intrinsics, now only the key schedule is in assembly...

Replace .byte hard coded instructions w/ the proper instructions now
that both clang and gcc support them...

On my machine, pulling the code to userland I saw performance go from
~150MB/sec to 2GB/sec in XTS mode.  GELI on GNOP saw a more modest
increase of about 3x due to other system overhead (geom and
opencrypto)...

These changes allow almost full disk io rate w/ geli...

Reviewed by:	-current, -security
Thanks to:	Mike Hamburg for the XTS tweek algorithm
2013-09-03 18:31:23 +00:00
andre
01fc9682ce Fix const propagation issues to make GCC happy.
Submitted by:	Michael Butler <imb@protected-networks.net>
2013-07-11 16:27:11 +00:00
andre
5c4a796865 SipHash is a cryptographically strong pseudo-random function (a.k.a. keyed
hash function) optimized for speed on short messages returning a 64bit hash/
digest value.

SipHash is simpler and much faster than other secure MACs and competitive
in speed with popular non-cryptographic hash functions.  It uses a 128-bit
key without the hidden cost of a key expansion step.  SipHash iterates a
simple round function consisting of four additions, four xors, and six
rotations, interleaved with xors of message blocks for a pre-defined number
of compression and finalization rounds.  The absence of  secret load/store
addresses or secret branch conditions avoid timing attacks.  No state is
shared between messages.  Hashing is deterministic and doesn't use nonces.
It is not susceptible to length extension attacks.

Target applications include network traffic authentication, message
authentication (MAC) and hash-tables protection against hash-flooding
denial-of-service attacks.

The number of update/finalization rounds is defined during initialization:

 SipHash24_Init() for the fast and reasonable strong version.
 SipHash48_Init() for the strong version (half as fast).

SipHash usage is similar to other hash functions:

 struct SIPHASH_CTX ctx;
 char *k = "16bytes long key"
 char *s = "string";
 uint64_t h = 0;
 SipHash24_Init(&ctx);
 SipHash_SetKey(&ctx, k);
 SipHash_Update(&ctx, s, strlen(s));
 SipHash_Final(&h, &ctx);  /* or */
 h = SipHash_End(&ctx);    /* or */
 h = SipHash24(&ctx, k, s, strlen(s));

It was designed by Jean-Philippe Aumasson and Daniel J. Bernstein and
is described in the paper "SipHash: a fast short-input PRF", 2012.09.18:
 https://131002.net/siphash/siphash.pdf
 Permanent ID: b9a943a805fbfc6fde808af9fc0ecdfa

Implemented by:	andre (based on the paper)
Reviewed by:	cperciva
2013-07-11 14:18:38 +00:00
delphij
b8b7a65ef6 Sync with KAME.
MFC after:	1 month
2013-07-09 22:04:35 +00:00
rmh
7e76036b6d Allow assert() to operate correctly when building userland code. 2013-07-09 10:27:26 +00:00
pjd
2d6cf6d680 When porting XTS-related code from OpenBSD I forgot to update copyright (only
OpenBSD was credited in one of two commits). Fix it.

Reported by:	Theo de Raadt <deraadt@cvs.openbsd.org>
Reviewed by:	Damien Miller <djm@mindrot.org>
2013-02-20 22:59:53 +00:00
kevlo
ceb08698f2 Revert previous commit...
Pointyhat to:	kevlo (myself)
2012-10-10 08:36:38 +00:00
kevlo
8747a46991 Prefer NULL over 0 for pointers 2012-10-09 08:27:40 +00:00
kib
361bfae5c2 Add support for the extended FPU states on amd64, both for native
64bit and 32bit ABIs.  As a side-effect, it enables AVX on capable
CPUs.

In particular:

- Query the CPU support for XSAVE, list of the supported extensions
  and the required size of FPU save area. The hw.use_xsave tunable is
  provided for disabling XSAVE, and hw.xsave_mask may be used to
  select the enabled extensions.

- Remove the FPU save area from PCB and dynamically allocate the
  (run-time sized) user save area on the top of the kernel stack,
  right above the PCB. Reorganize the thread0 PCB initialization to
  postpone it after BSP is queried for save area size.

- The dumppcb, stoppcbs and susppcbs now do not carry the FPU state as
  well. FPU state is only useful for suspend, where it is saved in
  dynamically allocated suspfpusave area.

- Use XSAVE and XRSTOR to save/restore FPU state, if supported and
  enabled.

- Define new mcontext_t flag _MC_HASFPXSTATE, indicating that
  mcontext_t has a valid pointer to out-of-struct extended FPU
  state. Signal handlers are supplied with stack-allocated fpu
  state. The sigreturn(2) and setcontext(2) syscall honour the flag,
  allowing the signal handlers to inspect and manipilate extended
  state in the interrupted context.

- The getcontext(2) never returns extended state, since there is no
  place in the fixed-sized mcontext_t to place variable-sized save
  area. And, since mcontext_t is embedded into ucontext_t, makes it
  impossible to fix in a reasonable way.  Instead of extending
  getcontext(2) syscall, provide a sysarch(2) facility to query
  extended FPU state.

- Add ptrace(2) support for getting and setting extended state; while
  there, implement missed PT_I386_{GET,SET}XMMREGS for 32bit binaries.

- Change fpu_kern KPI to not expose struct fpu_kern_ctx layout to
  consumers, making it opaque. Internally, struct fpu_kern_ctx now
  contains a space for the extended state. Convert in-kernel consumers
  of fpu_kern KPI both on i386 and amd64.

First version of the support for AVX was submitted by Tim Bird
<tim.bird am sony com> on behalf of Sony. This version was written
from scratch.

Tested by:	pho (previous version), Yamagi Burmeister <lists yamagi org>
MFC after:	1 month
2012-01-21 17:45:27 +00:00
pjd
0c79680192 Update Copyright.
MFC after:	3 days
2011-10-27 14:15:26 +00:00
pjd
d01d35973a Improve AES-NI performance for AES-XTS:
- Operate on uint64_t types when doing XORing, etc. instead of uint8_t.
- Don't bzero() temporary block for every AES block. Do it once for entire
  data block.
- AES-NI is available only on little endian architectures. Simplify code
  that takes block number from IV.

Benchmarks:

Memory-backed md(4) device, software AES-XTS, 4kB sector:

	# dd if=/dev/md0.eli bs=1m
	59.61MB/s

Memory-backed md(4) device, old AES-NI AES-XTS, 4kB sector:

	# dd if=/dev/md0.eli bs=1m
	97.29MB/s

Memory-backed md(4) device, new AES-NI AES-XTS, 4kB sector:

	# dd if=/dev/md0.eli bs=1m
	221.26MB/s

127% performance improvement between old and new code.

Harddisk, raw speed:

	# dd if=/dev/ada0 bs=1m
	137.63MB/s

Harddisk, software AES-XTS, 4kB sector:

	# dd if=/dev/ada0.eli bs=1m
	47.83MB/s (34% of raw disk speed)

Harddisk, old AES-NI AES-XTS, 4kB sector:

	# dd if=/dev/ada0.eli bs=1m
	68.33MB/s (49% of raw disk speed)

Harddisk, new AES-NI AES-XTS, 4kB sector:

	# dd if=/dev/ada0.eli bs=1m
	108.35MB/s (78% of raw disk speed)

58% performance improvement between old and new code.

As a side-note, GELI with AES-NI using AES-CBC can achive native disk speed.

MFC after:	3 days
2011-10-27 14:07:57 +00:00
kib
3ba18a4657 Fix a bug in the result of manual assembly.
Reported by:	Stefan Grundmann <sg2342 googlemail com>
PR:	kern/155118
MFC after:	3 days
2011-03-02 14:56:58 +00:00
brucec
2f4588c022 Make private functions static.
PR:		kern/43611
Submitted by:	Matt Emmerton <matt at gsicomp.on.ca>
Reviewed by:	kib
MFC after:	3 days
2011-02-21 16:21:43 +00:00
kib
1ddadd8390 Remove DEBUG sections.
MFC after:	3 days
2010-11-27 15:41:44 +00:00
kib
b2984c404f MFaesni r215427:
Only save FPU context when not executing in the context of the crypto
thread.

Tested by:	Mike Tancsa
MFC after:	1 week
2010-11-26 14:35:20 +00:00
kib
96d3f81fc8 Only save FPU context when not executing in the context of the crypto
thread.

Tested by:	Mike Tancsa
2010-11-17 16:17:15 +00:00
dim
2cbf038dc7 Change two missed instances of 'retq' in aeskeys_i386.S to 'retl', which
makes it possible to assemble this file with gas from newer binutils.

Reviewed by:	kib
2010-10-13 17:55:53 +00:00
pjd
5afa0ed148 Fix two copy&paste bugs.
MFC after:	2 weeks
2010-09-25 10:32:52 +00:00
pjd
419759c089 Add support for CRYPTO_AES_XTS.
MFC after:	1 week
2010-09-23 11:57:25 +00:00
pjd
7c5b734f64 Add support for CRD_F_KEY_EXPLICIT flag.
MFC after:	1 week
2010-09-23 11:46:53 +00:00
pjd
4c64468739 Simplify code a bit.
MFC after:	1 week
2010-09-23 11:26:38 +00:00
kib
a619846587 Crypto(4) driver for AESNI.
The aeskeys_{amd64,i386}.S content was mostly obtained from OpenBSD,
no objections to the license from core.

Hardware provided by:	Sentex Communications
Tested by:	fabient, pho (previous versions)
MFC after:	1 month
2010-07-23 11:00:46 +00:00
kib
6a41cd83d7 Use the fpu_kern_enter() interface to properly separate usermode FPU
context from in-kernel execution of padlock instructions and to handle
spurious FPUDNA exceptions that sometime are raised when doing padlock
calculations.

Globally mark crypto(9) kthread as using FPU.

Reviewed by:	pjd
Hardware provided by:	Sentex Communications
Tested by:	  pho
PR:    amd64/135014
MFC after:    1 month
2010-06-05 16:00:53 +00:00
vanhu
f484650153 Changed to M_NOWAIT when reallocing psc_buf in padlock_sha_update(),
as we already hold the non sleepable crypto_driver_mutex.

Approved by:	gnn(mentor)
Obtained from:	NETASQ
MFC after:	2 weeks
2009-05-27 09:52:12 +00:00
imp
6388654b8d identify routine takes driver_t *, not device_t *. 2009-02-05 19:30:28 +00:00
jkim
926a84447a Connect padlock(4) to amd64 build for VIA Nano processors. 2009-01-12 19:23:46 +00:00
mav
f702951b49 Avoid 256 integer divisions per rc4_init() call. Replace it with using
separate index variable.

It gives more then double rc4_init() performance increase on tested i386 P4.
It also gives about 15% speedup to PPTP VPN with stateless MPPE encryption
(by ng_mppc) which calls rc4_init() for every packet.
2008-12-16 13:58:37 +00:00
philip
f3d74d62d3 Fix a potential NULL-pointer dereference in padlock(4).
Spotted by:	Coverity (via pjd)
MFC after:	1 week
2008-11-17 19:00:36 +00:00
pjd
dca2fd863d Simplify session selection/allocation. 2008-08-09 20:01:01 +00:00
pjd
c2715b68b6 - Fix freeing session on newsession failure.
- Update copyright years.
2008-08-09 19:48:59 +00:00
pjd
e3f8d38bb0 Implify sessions freeing loop. 2008-08-09 19:47:19 +00:00