Commit Graph

32 Commits

Author SHA1 Message Date
Robert Watson
5e3f7694b1 Replace custom file descriptor array sleep lock constructed using a mutex
and flags with an sxlock.  This leads to a significant and measurable
performance improvement as a result of access to shared locking for
frequent lookup operations, reduced general overhead, and reduced overhead
in the event of contention.  All of these are imported for threaded
applications where simultaneous access to a shared file descriptor array
occurs frequently.  Kris has reported 2x-4x transaction rate improvements
on 8-core MySQL benchmarks; smaller improvements can be expected for many
workloads as a result of reduced overhead.

- Generally eliminate the distinction between "fast" and regular
  acquisisition of the filedesc lock; the plan is that they will now all
  be fast.  Change all locking instances to either shared or exclusive
  locks.

- Correct a bug (pointed out by kib) in fdfree() where previously msleep()
  was called without the mutex held; sx_sleep() is now always called with
  the sxlock held exclusively.

- Universally hold the struct file lock over changes to struct file,
  rather than the filedesc lock or no lock.  Always update the f_ops
  field last. A further memory barrier is required here in the future
  (discussed with jhb).

- Improve locking and reference management in linux_at(), which fails to
  properly acquire vnode references before using vnode pointers.  Annotate
  improper use of vn_fullpath(), which will be replaced at a future date.

In fcntl(), we conservatively acquire an exclusive lock, even though in
some cases a shared lock may be sufficient, which should be revisited.
The dropping of the filedesc lock in fdgrowtable() is no longer required
as the sxlock can be held over the sleep operation; we should consider
removing that (pointed out by attilio).

Tested by:	kris
Discussed with:	jhb, kris, attilio, jeff
2007-04-04 09:11:34 +00:00
Sam Leffler
6810ad6f2a Overhaul driver/subsystem api's:
o make all crypto drivers have a device_t; pseudo drivers like the s/w
  crypto driver synthesize one
o change the api between the crypto subsystem and drivers to use kobj;
  cryptodev_if.m defines this api
o use the fact that all crypto drivers now have a device_t to add support
  for specifying which of several potential devices to use when doing
  crypto operations
o add new ioctls that allow user apps to select a specific crypto device
  to use (previous ioctls maintained for compatibility)
o overhaul crypto subsystem code to eliminate lots of cruft and hide
  implementation details from drivers
o bring in numerous fixes from Michale Richardson/hifn; mostly for
  795x parts
o add an optional mechanism for mmap'ing the hifn 795x public key h/w
  to user space for use by openssl (not enabled by default)
o update crypto test tools to use new ioctl's and add cmd line options
  to specify a device to use for tests

These changes will also enable much future work on improving the core
crypto subsystem; including proper load balancing and interposing code
between the core and drivers to dispatch small operations to the s/w
driver as appropriate.

These changes were instigated by the work of Michael Richardson.

Reviewed by:	pjd
Approved by:	re
2007-03-21 03:42:51 +00:00
Pawel Jakub Dawidek
cd80523efc Fix usage of HMAC algorithms via /dev/crypto. 2006-05-22 16:24:11 +00:00
Pawel Jakub Dawidek
f6c4bc3b91 - Fix a very old bug in HMAC/SHA{384,512}. When HMAC is using SHA384
or SHA512, the blocksize is 128 bytes, not 64 bytes as anywhere else.
  The bug also exists in NetBSD, OpenBSD and various other independed
  implementations I look at.
- We cannot decide which hash function to use for HMAC based on the key
  length, because any HMAC function can use any key length.
  To fix it split CRYPTO_SHA2_HMAC into three algorithm:
  CRYPTO_SHA2_256_HMAC, CRYPTO_SHA2_384_HMAC and CRYPTO_SHA2_512_HMAC.
  Those names are consistent with OpenBSD's naming.
- Remove authsize field from auth_hash structure.
- Allow consumer to define size of hash he wants to receive.
  This allows to use HMAC not only for IPsec, where 96 bits MAC is requested.
  The size of requested MAC is defined at newsession time in the cri_mlen
  field - when 0, entire MAC will be returned.
- Add swcr_authprepare() function which prepares authentication key.
- Allow to provide key for every authentication operation, not only at
  newsession time by honoring CRD_F_KEY_EXPLICIT flag.
- Make giving key at newsession time optional - don't try to operate on it
  if its NULL.
- Extend COPYBACK()/COPYDATA() macros to handle CRYPTO_BUF_CONTIG buffer
  type as well.
- Accept CRYPTO_BUF_IOV buffer type in swcr_authcompute() as we have
  cuio_apply() now.
- 16 bits for key length (SW_klen) is more than enough.

Reviewed by:	sam
2006-05-17 18:24:17 +00:00
Pawel Jakub Dawidek
613894d047 If kern.cryptodevallowsoft is TRUE allow also for symmetric software crypto
in kernel. Useful for testing.

Reviewed by:	sam
2006-05-17 18:01:51 +00:00
Wojciech A. Koszek
51b4ccb464 This patch fixes a problem, which exists if you have IPSEC in your kernel
and want to have crypto support loaded as KLD. By moving zlib to separate
module and adding MODULE_DEPEND directives, it is possible to use such
configuration without complication. Otherwise, since IPSEC is linked with
zlib (just like crypto.ko) you'll get following error:

	interface zlib.1 already present in the KLD 'kernel'!

Approved by:	cognet (mentor)
2006-02-27 16:56:22 +00:00
Pawel Jakub Dawidek
e6d944d7c3 Fix bogus check. It was possible to panic the kernel by giving 0 length.
This is actually a local DoS, as every user can use /dev/crypto if there
is crypto hardware in the system and cryptodev.ko is loaded (or compiled
into the kernel).

Reported by:	Mike Tancsa <mike@sentex.net>
MFC after:	1 day
2005-08-18 11:58:03 +00:00
Poul-Henning Kamp
78b7c8d68d Use dynamic major number allocation. 2005-02-27 22:11:02 +00:00
Warner Losh
60727d8b86 /* -> /*- for license, minor formatting changes 2005-01-07 02:29:27 +00:00
Poul-Henning Kamp
a0fbccc9e7 Push Giant down through ioctl.
Don't grab Giant in the upper syscall/wrapper code

NET_LOCK_GIANT in the socket code (sockets/fifos).

mtx_lock(&Giant) in the vnode code.

mtx_lock(&Giant) in the opencrypto code.  (This may actually not be
needed, but better safe than sorry).

Devfs grabs Giant if the driver is marked as needing Giant.
2004-11-17 09:09:55 +00:00
Robert Watson
d7aed12f45 Don't acquire Giant in cryptof_close(), as the code is intended to be
able to run MPsafe (and appears to be MPsafe).

Discussed with (some time ago):	sam
2004-08-10 03:26:17 +00:00
Robert Watson
1c1ce9253f Push acquisition of Giant from fdrop_closed() into fo_close() so that
individual file object implementations can optionally acquire Giant if
they require it:

- soo_close(): depends on debug.mpsafenet
- pipe_close(): Giant not acquired
- kqueue_close(): Giant required
- vn_close(): Giant required
- cryptof_close(): Giant required (conservative)

Notes:

  Giant is still acquired in close() even when closing MPSAFE objects
  due to kqueue requiring Giant in the calling closef() code.
  Microbenchmarks indicate that this removal of Giant cuts 3%-3% off
  of pipe create/destroy pairs from user space with SMP compiled into
  the kernel.

  The cryptodev and opencrypto code appears MPSAFE, but I'm unable to
  test it extensively and so have left Giant over fo_close().  It can
  probably be removed given some testing and review.
2004-07-22 18:35:43 +00:00
Poul-Henning Kamp
89c9c53da0 Do the dreaded s/dev_t/struct cdev */
Bump __FreeBSD_version accordingly.
2004-06-16 09:47:26 +00:00
Poul-Henning Kamp
5dba30f15a add missing #include <sys/module.h> 2004-05-30 20:27:19 +00:00
Poul-Henning Kamp
dc08ffec87 Device megapatch 4/6:
Introduce d_version field in struct cdevsw, this must always be
initialized to D_VERSION.

Flip sense of D_NOGIANT flag to D_NEEDGIANT, this involves removing
four D_NOGIANT flags and adding 145 D_NEEDGIANT flags.
2004-02-21 21:10:55 +00:00
Sam Leffler
57053a10cf eliminate an unnecessary 8Kbyte bzero that was being done for each
submitted operation

Submitted by:	Thor Lancelot Simon
Reviewed by:	jhb
Approved by:	re (jhb)
2003-11-19 22:42:34 +00:00
David Malone
e1419c08e2 falloc allocates a file structure and adds it to the file descriptor
table, acquiring the necessary locks as it works. It usually returns
two references to the new descriptor: one in the descriptor table
and one via a pointer argument.

As falloc releases the FILEDESC lock before returning, there is a
potential for a process to close the reference in the file descriptor
table before falloc's caller gets to use the file. I don't think this
can happen in practice at the moment, because Giant indirectly protects
closes.

To stop the file being completly closed in this situation, this change
makes falloc set the refcount to two when both references are returned.
This makes life easier for several of falloc's callers, because the
first thing they previously did was grab an extra reference on the
file.

Reviewed by:	iedowse
Idea run past:	jhb
2003-10-19 20:41:07 +00:00
Poul-Henning Kamp
70cd771337 The present defaults for the open and close for device drivers which
provide no methods does not make any sense, and is not used by any
driver.

It is a pretty hard to come up with even a theoretical concept of
a device driver which would always fail open and close with ENODEV.

Change the defaults to be nullopen() and nullclose() which simply
does nothing.

Remove explicit initializations to these from the drivers which
already used them.
2003-09-27 12:01:01 +00:00
Poul-Henning Kamp
7c2d2efd58 Initialize struct fileops with C99 sparse initialization. 2003-06-18 18:16:40 +00:00
David E. O'Brien
2c44651495 Use __FBSDID(). 2003-06-11 05:57:50 +00:00
Sam Leffler
3569ae7f66 Flush my local cache of cryto subsystem fixes:
o add a ``done'' flag for crypto operations; this is set when the operation
  completes and is intended for callers to check operations that may complete
  ``prematurely'' because of direct callbacks
o close a race for operations where the crypto driver returns ERESTART: we
  need to hold the q lock to insure the blocked state for the driver and any
  driver-private state is consistent; otherwise drivers may take an interrupt
  and notify the crypto subsystem that it can unblock the driver but operations
  will be left queued and never be processed
o close a race in /dev/crypto where operations can complete before the caller
  can sleep waiting for the callback: use a per-session mutex and the new done
  flag to handle this
o correct crypto_dispatch's handling of operations where the driver returns
  ERESTART: the return value must be zero and not ERESTART, otherwise the
  caller may free the crypto request despite it being queued for later handling
  (this typically results in a later panic)
o change crypto mutex ``names'' so witness printouts and the like are more
  meaningful
2003-06-02 23:28:06 +00:00
Poul-Henning Kamp
7ac40f5f59 Gigacommit to improve device-driver source compatibility between
branches:

Initialize struct cdevsw using C99 sparse initializtion and remove
all initializations to default values.

This patch is automatically generated and has been tested by compiling
LINT with all the fields in struct cdevsw in reverse order on alpha,
sparc64 and i386.

Approved by:    re(scottl)
2003-03-03 12:15:54 +00:00
Sam Leffler
eb73a605cd o add a CRYPTO_F_CBIMM flag to symmetric ops to indicate the callback
should be done in crypto_done rather than in the callback thread
o use this flag to mark operations from /dev/crypto since the callback
  routine just does a wakeup; this eliminates the last unneeded ctx switch
o change CRYPTO_F_NODELAY to CRYPTO_F_BATCH with an inverted meaning
  so "0" becomes the default/desired setting (needed for user-mode
  compatibility with openbsd)
o change crypto_dispatch to honor CRYPTO_F_BATCH instead of always
  dispatching immediately
o remove uses of CRYPTO_F_NODELAY
o define COP_F_BATCH for ops submitted through /dev/crypto and pass
  this on to the op that is submitted

Similar changes and more eventually coming for asymmetric ops.

MFC if re gives approval.
2003-02-23 07:25:48 +00:00
Warner Losh
a163d034fa Back out M_* changes, per decision of the TRB.
Approved by: trb
2003-02-19 05:47:46 +00:00
Alfred Perlstein
44956c9863 Remove M_TRYWAIT/M_WAITOK/M_WAIT. Callers should use 0.
Merge M_NOWAIT/M_DONTWAIT into a single flag M_NOWAIT.
2003-01-21 08:56:16 +00:00
Matthew Dillon
48e3128b34 Bow to the whining masses and change a union back into void *. Retain
removal of unnecessary casts and throw in some minor cleanups to see if
anyone complains, just for the hell of it.
2003-01-13 00:33:17 +00:00
Matthew Dillon
cd72f2180b Change struct file f_data to un_data, a union of the correct struct
pointer types, and remove a huge number of casts from code using it.

Change struct xfile xf_data to xun_data (ABI is still compatible).

If we need to add a #define for f_data and xf_data we can, but I don't
think it will be necessary.  There are no operational changes in this
commit.
2003-01-12 01:37:13 +00:00
Sam Leffler
7d1853ee0e MFS: crypto timing support; purge usercrypto sysctl (just don't config
cryptodev or kldunload cryptodev module); crypto statistcs; remove
unused alloctype field from crypto op to offset addition of the
performance time stamp

Supported by:	Vernier Networks
2003-01-03 06:16:59 +00:00
Sam Leffler
955630483b correct minor # in make_dev call
Submitted by:	Doug Ambrisko" <ambrisko@verniernetworks.com>
2002-11-08 23:07:41 +00:00
Mark Murray
f544a52873 Module-ize the 'core' crypto stuff. This may still need to be compiled
into the kernel by default (if required), but other modules can now
depend() on this.

Fix inter-module dependancy.

Earlier version OK'ed by:	sam
2002-10-16 14:31:34 +00:00
Sam Leffler
b4c408e0fe remove CIOGSSESSION (get software session); it was added only for testing 2002-10-07 18:37:31 +00:00
Sam Leffler
091d81d134 In-kernel crypto framework derived from openbsd. This facility provides
a consistent interface to h/w and s/w crypto algorithms for use by the
kernel and (for h/w at least) by user-mode apps.  Access for user-level
code is through a /dev/crypto device that'll eventually be used by openssl
to (potentially) accelerate many applications.  Coming soon is an IPsec
that makes use of this service to accelerate ESP, AH, and IPCOMP protocols.

Included here is the "core" crypto support, /dev/crypto driver, various
crypto algorithms that are not already present in the KAME crypto area,
and support routines used by crypto device drivers.

Obtained from:	openbsd
2002-10-04 20:31:23 +00:00