freebsd-nq/sys/geom/eli/g_eli.c

1456 lines
40 KiB
C
Raw Normal View History

/*-
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
*
* Copyright (c) 2005-2019 Pawel Jakub Dawidek <pawel@dawidek.net>
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
2006-02-01 12:06:01 +00:00
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/cons.h>
#include <sys/kernel.h>
#include <sys/linker.h>
#include <sys/module.h>
#include <sys/lock.h>
#include <sys/mutex.h>
#include <sys/bio.h>
#include <sys/sbuf.h>
#include <sys/sysctl.h>
#include <sys/malloc.h>
#include <sys/eventhandler.h>
#include <sys/kthread.h>
#include <sys/proc.h>
#include <sys/sched.h>
#include <sys/smp.h>
#include <sys/uio.h>
#include <sys/vnode.h>
#include <vm/uma.h>
#include <geom/geom.h>
#include <geom/geom_dbg.h>
#include <geom/eli/g_eli.h>
#include <geom/eli/pkcs5v2.h>
#include <crypto/intake.h>
FEATURE(geom_eli, "GEOM crypto module");
MALLOC_DEFINE(M_ELI, "eli data", "GEOM_ELI Data");
SYSCTL_DECL(_kern_geom);
SYSCTL_NODE(_kern_geom, OID_AUTO, eli, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
"GEOM_ELI stuff");
static int g_eli_version = G_ELI_VERSION;
SYSCTL_INT(_kern_geom_eli, OID_AUTO, version, CTLFLAG_RD, &g_eli_version, 0,
"GELI version");
int g_eli_debug = 0;
SYSCTL_INT(_kern_geom_eli, OID_AUTO, debug, CTLFLAG_RWTUN, &g_eli_debug, 0,
"Debug level");
static u_int g_eli_tries = 3;
SYSCTL_UINT(_kern_geom_eli, OID_AUTO, tries, CTLFLAG_RWTUN, &g_eli_tries, 0,
"Number of tries for entering the passphrase");
static u_int g_eli_visible_passphrase = GETS_NOECHO;
SYSCTL_UINT(_kern_geom_eli, OID_AUTO, visible_passphrase, CTLFLAG_RWTUN,
&g_eli_visible_passphrase, 0,
"Visibility of passphrase prompt (0 = invisible, 1 = visible, 2 = asterisk)");
u_int g_eli_overwrites = G_ELI_OVERWRITES;
SYSCTL_UINT(_kern_geom_eli, OID_AUTO, overwrites, CTLFLAG_RWTUN, &g_eli_overwrites,
0, "Number of times on-disk keys should be overwritten when destroying them");
static u_int g_eli_threads = 0;
SYSCTL_UINT(_kern_geom_eli, OID_AUTO, threads, CTLFLAG_RWTUN, &g_eli_threads, 0,
"Number of threads doing crypto work");
u_int g_eli_batch = 0;
SYSCTL_UINT(_kern_geom_eli, OID_AUTO, batch, CTLFLAG_RWTUN, &g_eli_batch, 0,
"Use crypto operations batching");
/*
* Passphrase cached during boot, in order to be more user-friendly if
* there are multiple providers using the same passphrase.
*/
static char cached_passphrase[256];
static u_int g_eli_boot_passcache = 1;
TUNABLE_INT("kern.geom.eli.boot_passcache", &g_eli_boot_passcache);
SYSCTL_UINT(_kern_geom_eli, OID_AUTO, boot_passcache, CTLFLAG_RD,
&g_eli_boot_passcache, 0,
"Passphrases are cached during boot process for possible reuse");
static void
fetch_loader_passphrase(void * dummy)
{
char * env_passphrase;
KASSERT(dynamic_kenv, ("need dynamic kenv"));
if ((env_passphrase = kern_getenv("kern.geom.eli.passphrase")) != NULL) {
/* Extract passphrase from the environment. */
strlcpy(cached_passphrase, env_passphrase,
sizeof(cached_passphrase));
freeenv(env_passphrase);
/* Wipe the passphrase from the environment. */
kern_unsetenv("kern.geom.eli.passphrase");
}
}
SYSINIT(geli_fetch_loader_passphrase, SI_SUB_KMEM + 1, SI_ORDER_ANY,
fetch_loader_passphrase, NULL);
static void
zero_boot_passcache(void)
{
explicit_bzero(cached_passphrase, sizeof(cached_passphrase));
}
static void
zero_geli_intake_keys(void)
{
struct keybuf *keybuf;
int i;
if ((keybuf = get_keybuf()) != NULL) {
/* Scan the key buffer, clear all GELI keys. */
for (i = 0; i < keybuf->kb_nents; i++) {
if (keybuf->kb_ents[i].ke_type == KEYBUF_TYPE_GELI) {
explicit_bzero(keybuf->kb_ents[i].ke_data,
sizeof(keybuf->kb_ents[i].ke_data));
keybuf->kb_ents[i].ke_type = KEYBUF_TYPE_NONE;
}
}
}
}
static void
zero_intake_passcache(void *dummy)
{
zero_boot_passcache();
zero_geli_intake_keys();
}
EVENTHANDLER_DEFINE(mountroot, zero_intake_passcache, NULL, 0);
static eventhandler_tag g_eli_pre_sync = NULL;
static int g_eli_read_metadata_offset(struct g_class *mp, struct g_provider *pp,
off_t offset, struct g_eli_metadata *md);
static int g_eli_destroy_geom(struct gctl_req *req, struct g_class *mp,
struct g_geom *gp);
static void g_eli_init(struct g_class *mp);
static void g_eli_fini(struct g_class *mp);
static g_taste_t g_eli_taste;
static g_dumpconf_t g_eli_dumpconf;
struct g_class g_eli_class = {
.name = G_ELI_CLASS_NAME,
.version = G_VERSION,
.ctlreq = g_eli_config,
.taste = g_eli_taste,
.destroy_geom = g_eli_destroy_geom,
.init = g_eli_init,
.fini = g_eli_fini
};
/*
* Code paths:
* BIO_READ:
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
* g_eli_start -> g_eli_crypto_read -> g_io_request -> g_eli_read_done -> g_eli_crypto_run -> g_eli_crypto_read_done -> g_io_deliver
* BIO_WRITE:
* g_eli_start -> g_eli_crypto_run -> g_eli_crypto_write_done -> g_io_request -> g_eli_write_done -> g_io_deliver
*/
/*
* EAGAIN from crypto(9) means, that we were probably balanced to another crypto
* accelerator or something like this.
* The function updates the SID and rerun the operation.
*/
int
g_eli_crypto_rerun(struct cryptop *crp)
{
struct g_eli_softc *sc;
struct g_eli_worker *wr;
struct bio *bp;
int error;
bp = (struct bio *)crp->crp_opaque;
sc = bp->bio_to->geom->softc;
LIST_FOREACH(wr, &sc->sc_workers, w_next) {
if (wr->w_number == bp->bio_pflags)
break;
}
KASSERT(wr != NULL, ("Invalid worker (%u).", bp->bio_pflags));
G_ELI_DEBUG(1, "Rerunning crypto %s request (sid: %p -> %p).",
bp->bio_cmd == BIO_READ ? "READ" : "WRITE", wr->w_sid,
crp->crp_session);
wr->w_sid = crp->crp_session;
crp->crp_etype = 0;
error = crypto_dispatch(crp);
if (error == 0)
return (0);
G_ELI_DEBUG(1, "%s: crypto_dispatch() returned %d.", __func__, error);
crp->crp_etype = error;
return (error);
}
static void
g_eli_getattr_done(struct bio *bp)
{
if (bp->bio_error == 0 &&
!strcmp(bp->bio_attribute, "GEOM::physpath")) {
strlcat(bp->bio_data, "/eli", bp->bio_length);
}
g_std_done(bp);
}
/*
* The function is called afer reading encrypted data from the provider.
*
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
* g_eli_start -> g_eli_crypto_read -> g_io_request -> G_ELI_READ_DONE -> g_eli_crypto_run -> g_eli_crypto_read_done -> g_io_deliver
*/
void
g_eli_read_done(struct bio *bp)
{
struct g_eli_softc *sc;
struct bio *pbp;
G_ELI_LOGREQ(2, bp, "Request done.");
pbp = bp->bio_parent;
if (pbp->bio_error == 0 && bp->bio_error != 0)
pbp->bio_error = bp->bio_error;
g_destroy_bio(bp);
/*
* Do we have all sectors already?
*/
pbp->bio_inbed++;
if (pbp->bio_inbed < pbp->bio_children)
return;
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
sc = pbp->bio_to->geom->softc;
if (pbp->bio_error != 0) {
G_ELI_LOGREQ(0, pbp, "%s() failed (error=%d)", __func__,
pbp->bio_error);
pbp->bio_completed = 0;
if (pbp->bio_driver2 != NULL) {
free(pbp->bio_driver2, M_ELI);
pbp->bio_driver2 = NULL;
}
g_io_deliver(pbp, pbp->bio_error);
if (sc != NULL)
atomic_subtract_int(&sc->sc_inflight, 1);
return;
}
mtx_lock(&sc->sc_queue_mtx);
bioq_insert_tail(&sc->sc_queue, pbp);
mtx_unlock(&sc->sc_queue_mtx);
wakeup(sc);
}
/*
* The function is called after we encrypt and write data.
*
* g_eli_start -> g_eli_crypto_run -> g_eli_crypto_write_done -> g_io_request -> G_ELI_WRITE_DONE -> g_io_deliver
*/
void
g_eli_write_done(struct bio *bp)
{
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
struct g_eli_softc *sc;
struct bio *pbp;
G_ELI_LOGREQ(2, bp, "Request done.");
pbp = bp->bio_parent;
if (pbp->bio_error == 0 && bp->bio_error != 0)
pbp->bio_error = bp->bio_error;
g_destroy_bio(bp);
/*
* Do we have all sectors already?
*/
pbp->bio_inbed++;
if (pbp->bio_inbed < pbp->bio_children)
return;
free(pbp->bio_driver2, M_ELI);
pbp->bio_driver2 = NULL;
if (pbp->bio_error != 0) {
G_ELI_LOGREQ(0, pbp, "%s() failed (error=%d)", __func__,
pbp->bio_error);
pbp->bio_completed = 0;
} else
pbp->bio_completed = pbp->bio_length;
/*
* Write is finished, send it up.
*/
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
sc = pbp->bio_to->geom->softc;
g_io_deliver(pbp, pbp->bio_error);
if (sc != NULL)
atomic_subtract_int(&sc->sc_inflight, 1);
}
/*
* This function should never be called, but GEOM made as it set ->orphan()
* method for every geom.
*/
static void
g_eli_orphan_spoil_assert(struct g_consumer *cp)
{
panic("Function %s() called for %s.", __func__, cp->geom->name);
}
static void
g_eli_orphan(struct g_consumer *cp)
{
struct g_eli_softc *sc;
g_topology_assert();
sc = cp->geom->softc;
if (sc == NULL)
return;
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
g_eli_destroy(sc, TRUE);
}
static void
g_eli_resize(struct g_consumer *cp)
{
struct g_eli_softc *sc;
struct g_provider *epp, *pp;
off_t oldsize;
g_topology_assert();
sc = cp->geom->softc;
if (sc == NULL)
return;
if ((sc->sc_flags & G_ELI_FLAG_AUTORESIZE) == 0) {
G_ELI_DEBUG(0, "Autoresize is turned off, old size: %jd.",
(intmax_t)sc->sc_provsize);
return;
}
pp = cp->provider;
if ((sc->sc_flags & G_ELI_FLAG_ONETIME) == 0) {
struct g_eli_metadata md;
u_char *sector;
int error;
sector = NULL;
error = g_eli_read_metadata_offset(cp->geom->class, pp,
sc->sc_provsize - pp->sectorsize, &md);
if (error != 0) {
G_ELI_DEBUG(0, "Cannot read metadata from %s (error=%d).",
pp->name, error);
goto iofail;
}
md.md_provsize = pp->mediasize;
sector = malloc(pp->sectorsize, M_ELI, M_WAITOK | M_ZERO);
eli_metadata_encode(&md, sector);
error = g_write_data(cp, pp->mediasize - pp->sectorsize, sector,
pp->sectorsize);
if (error != 0) {
G_ELI_DEBUG(0, "Cannot store metadata on %s (error=%d).",
pp->name, error);
goto iofail;
}
explicit_bzero(sector, pp->sectorsize);
error = g_write_data(cp, sc->sc_provsize - pp->sectorsize,
sector, pp->sectorsize);
if (error != 0) {
G_ELI_DEBUG(0, "Cannot clear old metadata from %s (error=%d).",
pp->name, error);
goto iofail;
}
iofail:
explicit_bzero(&md, sizeof(md));
zfree(sector, M_ELI);
}
oldsize = sc->sc_mediasize;
sc->sc_mediasize = eli_mediasize(sc, pp->mediasize, pp->sectorsize);
g_eli_key_resize(sc);
sc->sc_provsize = pp->mediasize;
epp = LIST_FIRST(&sc->sc_geom->provider);
g_resize_provider(epp, sc->sc_mediasize);
G_ELI_DEBUG(0, "Device %s size changed from %jd to %jd.", epp->name,
(intmax_t)oldsize, (intmax_t)sc->sc_mediasize);
}
/*
* BIO_READ:
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
* G_ELI_START -> g_eli_crypto_read -> g_io_request -> g_eli_read_done -> g_eli_crypto_run -> g_eli_crypto_read_done -> g_io_deliver
* BIO_WRITE:
* G_ELI_START -> g_eli_crypto_run -> g_eli_crypto_write_done -> g_io_request -> g_eli_write_done -> g_io_deliver
*/
2006-02-01 12:06:01 +00:00
static void
g_eli_start(struct bio *bp)
2006-02-01 12:06:01 +00:00
{
struct g_eli_softc *sc;
struct g_consumer *cp;
struct bio *cbp;
sc = bp->bio_to->geom->softc;
KASSERT(sc != NULL,
("Provider's error should be set (error=%d)(device=%s).",
bp->bio_to->error, bp->bio_to->name));
G_ELI_LOGREQ(2, bp, "Request received.");
switch (bp->bio_cmd) {
case BIO_READ:
case BIO_WRITE:
case BIO_GETATTR:
case BIO_FLUSH:
Add support for managing Shingled Magnetic Recording (SMR) drives. This change includes support for SCSI SMR drives (which conform to the Zoned Block Commands or ZBC spec) and ATA SMR drives (which conform to the Zoned ATA Command Set or ZAC spec) behind SAS expanders. This includes full management support through the GEOM BIO interface, and through a new userland utility, zonectl(8), and through camcontrol(8). This is now ready for filesystems to use to detect and manage zoned drives. (There is no work in progress that I know of to use this for ZFS or UFS, if anyone is interested, let me know and I may have some suggestions.) Also, improve ATA command passthrough and dispatch support, both via ATA and ATA passthrough over SCSI. Also, add support to camcontrol(8) for the ATA Extended Power Conditions feature set. You can now manage ATA device power states, and set various idle time thresholds for a drive to enter lower power states. Note that this change cannot be MFCed in full, because it depends on changes to the struct bio API that break compatilibity. In order to avoid breaking the stable API, only changes that don't touch or depend on the struct bio changes can be merged. For example, the camcontrol(8) changes don't depend on the new bio API, but zonectl(8) and the probe changes to the da(4) and ada(4) drivers do depend on it. Also note that the SMR changes have not yet been tested with an actual SCSI ZBC device, or a SCSI to ATA translation layer (SAT) that supports ZBC to ZAC translation. I have not yet gotten a suitable drive or SAT layer, so any testing help would be appreciated. These changes have been tested with Seagate Host Aware SATA drives attached to both SAS and SATA controllers. Also, I do not have any SATA Host Managed devices, and I suspect that it may take additional (hopefully minor) changes to support them. Thanks to Seagate for supplying the test hardware and answering questions. sbin/camcontrol/Makefile: Add epc.c and zone.c. sbin/camcontrol/camcontrol.8: Document the zone and epc subcommands. sbin/camcontrol/camcontrol.c: Add the zone and epc subcommands. Add auxiliary register support to build_ata_cmd(). Make sure to set the CAM_ATAIO_NEEDRESULT, CAM_ATAIO_DMA, and CAM_ATAIO_FPDMA flags as appropriate for ATA commands. Add a new get_ata_status() function to parse ATA result from SCSI sense descriptors (for ATA passthrough over SCSI) and ATA I/O requests. sbin/camcontrol/camcontrol.h: Update the build_ata_cmd() prototype Add get_ata_status(), zone(), and epc(). sbin/camcontrol/epc.c: Support for ATA Extended Power Conditions features. This includes support for all features documented in the ACS-4 Revision 12 specification from t13.org (dated February 18, 2016). The EPC feature set allows putting a drive into a power power mode immediately, or setting timeouts so that the drive will automatically enter progressively lower power states after various idle times. sbin/camcontrol/fwdownload.c: Update the firmware download code for the new build_ata_cmd() arguments. sbin/camcontrol/zone.c: Implement support for Shingled Magnetic Recording (SMR) drives via SCSI Zoned Block Commands (ZBC) and ATA Zoned Device ATA Command Set (ZAC). These specs were developed in concert, and are functionally identical. The primary differences are due to SCSI and ATA differences. (SCSI is big endian, ATA is little endian, for example.) This includes support for all commands defined in the ZBC and ZAC specs. sys/cam/ata/ata_all.c: Decode a number of additional ATA command names in ata_op_string(). Add a new CCB building function, ata_read_log(). Add ata_zac_mgmt_in() and ata_zac_mgmt_out() CCB building functions. These support both DMA and NCQ encapsulation. sys/cam/ata/ata_all.h: Add prototypes for ata_read_log(), ata_zac_mgmt_out(), and ata_zac_mgmt_in(). sys/cam/ata/ata_da.c: Revamp the ada(4) driver to support zoned devices. Add four new probe states to gather information needed for zone support. Add a new adasetflags() function to avoid duplication of large blocks of flag setting between the async handler and register functions. Add new sysctl variables that describe zone support and paramters. Add support for the new BIO_ZONE bio, and all of its subcommands: DISK_ZONE_OPEN, DISK_ZONE_CLOSE, DISK_ZONE_FINISH, DISK_ZONE_RWP, DISK_ZONE_REPORT_ZONES, and DISK_ZONE_GET_PARAMS. sys/cam/scsi/scsi_all.c: Add command descriptions for the ZBC IN/OUT commands. Add descriptions for ZBC Host Managed devices. Add a new function, scsi_ata_pass() to do ATA passthrough over SCSI. This will eventually replace scsi_ata_pass_16() -- it can create the 12, 16, and 32-byte variants of the ATA PASS-THROUGH command, and supports setting all of the registers defined as of SAT-4, Revision 5 (March 11, 2016). Change scsi_ata_identify() to use scsi_ata_pass() instead of scsi_ata_pass_16(). Add a new scsi_ata_read_log() function to facilitate reading ATA logs via SCSI. sys/cam/scsi/scsi_all.h: Add the new ATA PASS-THROUGH(32) command CDB. Add extended and variable CDB opcodes. Add Zoned Block Device Characteristics VPD page. Add ATA Return SCSI sense descriptor. Add prototypes for scsi_ata_read_log() and scsi_ata_pass(). sys/cam/scsi/scsi_da.c: Revamp the da(4) driver to support zoned devices. Add five new probe states, four of which are needed for ATA devices. Add five new sysctl variables that describe zone support and parameters. The da(4) driver supports SCSI ZBC devices, as well as ATA ZAC devices when they are attached via a SCSI to ATA Translation (SAT) layer. Since ZBC -> ZAC translation is a new feature in the T10 SAT-4 spec, most SATA drives will be supported via ATA commands sent via the SCSI ATA PASS-THROUGH command. The da(4) driver will prefer the ZBC interface, if it is available, for performance reasons, but will use the ATA PASS-THROUGH interface to the ZAC command set if the SAT layer doesn't support translation yet. As I mentioned above, ZBC command support is untested. Add support for the new BIO_ZONE bio, and all of its subcommands: DISK_ZONE_OPEN, DISK_ZONE_CLOSE, DISK_ZONE_FINISH, DISK_ZONE_RWP, DISK_ZONE_REPORT_ZONES, and DISK_ZONE_GET_PARAMS. Add scsi_zbc_in() and scsi_zbc_out() CCB building functions. Add scsi_ata_zac_mgmt_out() and scsi_ata_zac_mgmt_in() CCB/CDB building functions. Note that these have return values, unlike almost all other CCB building functions in CAM. The reason is that they can fail, depending upon the particular combination of input parameters. The primary failure case is if the user wants NCQ, but fails to specify additional CDB storage. NCQ requires using the 32-byte version of the SCSI ATA PASS-THROUGH command, and the current CAM CDB size is 16 bytes. sys/cam/scsi/scsi_da.h: Add ZBC IN and ZBC OUT CDBs and opcodes. Add SCSI Report Zones data structures. Add scsi_zbc_in(), scsi_zbc_out(), scsi_ata_zac_mgmt_out(), and scsi_ata_zac_mgmt_in() prototypes. sys/dev/ahci/ahci.c: Fix SEND / RECEIVE FPDMA QUEUED in the ahci(4) driver. ahci_setup_fis() previously set the top bits of the sector count register in the FIS to 0 for FPDMA commands. This is okay for read and write, because the PRIO field is in the only thing in those bits, and we don't implement that further up the stack. But, for SEND and RECEIVE FPDMA QUEUED, the subcommand is in that byte, so it needs to be transmitted to the drive. In ahci_setup_fis(), always set the the top 8 bits of the sector count register. We need it in both the standard and NCQ / FPDMA cases. sys/geom/eli/g_eli.c: Pass BIO_ZONE commands through the GELI class. sys/geom/geom.h: Add g_io_zonecmd() prototype. sys/geom/geom_dev.c: Add new DIOCZONECMD ioctl, which allows sending zone commands to disks. sys/geom/geom_disk.c: Add support for BIO_ZONE commands. sys/geom/geom_disk.h: Add a new flag, DISKFLAG_CANZONE, that indicates that a given GEOM disk client can handle BIO_ZONE commands. sys/geom/geom_io.c: Add a new function, g_io_zonecmd(), that handles execution of BIO_ZONE commands. Add permissions check for BIO_ZONE commands. Add command decoding for BIO_ZONE commands. sys/geom/geom_subr.c: Add DDB command decoding for BIO_ZONE commands. sys/kern/subr_devstat.c: Record statistics for REPORT ZONES commands. Note that the number of bytes transferred for REPORT ZONES won't quite match what is received from the harware. This is because we're necessarily counting bytes coming from the da(4) / ada(4) drivers, which are using the disk_zone.h interface to communicate up the stack. The structure sizes it uses are slightly different than the SCSI and ATA structure sizes. sys/sys/ata.h: Add many bit and structure definitions for ZAC, NCQ, and EPC command support. sys/sys/bio.h: Convert the bio_cmd field to a straight enumeration. This will yield more space for additional commands in the future. After change r297955 and other related changes, this is now possible. Converting to an enumeration will also prevent use as a bitmask in the future. sys/sys/disk.h: Define the DIOCZONECMD ioctl. sys/sys/disk_zone.h: Add a new API for managing zoned disks. This is very close to the SCSI ZBC and ATA ZAC standards, but uses integers in native byte order instead of big endian (SCSI) or little endian (ATA) byte arrays. This is intended to offer to the complete feature set of the ZBC and ZAC disk management without requiring the application developer to include SCSI or ATA headers. We also use one set of headers for ioctl consumers and kernel bio-level consumers. sys/sys/param.h: Bump __FreeBSD_version for sys/bio.h command changes, and inclusion of SMR support. usr.sbin/Makefile: Add the zonectl utility. usr.sbin/diskinfo/diskinfo.c Add disk zoning capability to the 'diskinfo -v' output. usr.sbin/zonectl/Makefile: Add zonectl makefile. usr.sbin/zonectl/zonectl.8 zonectl(8) man page. usr.sbin/zonectl/zonectl.c The zonectl(8) utility. This allows managing SCSI or ATA zoned disks via the disk_zone.h API. You can report zones, reset write pointers, get parameters, etc. Sponsored by: Spectra Logic Differential Revision: https://reviews.freebsd.org/D6147 Reviewed by: wblock (documentation)
2016-05-19 14:08:36 +00:00
case BIO_ZONE:
case BIO_SPEEDUP:
break;
case BIO_DELETE:
/*
* If the user hasn't set the NODELETE flag, we just pass
* it down the stack and let the layers beneath us do (or
* not) whatever they do with it. If they have, we
* reject it. A possible extension would be an
* additional flag to take it as a hint to shred the data
* with [multiple?] overwrites.
*/
if (!(sc->sc_flags & G_ELI_FLAG_NODELETE))
break;
2006-09-30 08:16:49 +00:00
default:
g_io_deliver(bp, EOPNOTSUPP);
return;
}
cbp = g_clone_bio(bp);
if (cbp == NULL) {
g_io_deliver(bp, ENOMEM);
return;
}
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
bp->bio_driver1 = cbp;
bp->bio_pflags = G_ELI_NEW_BIO;
switch (bp->bio_cmd) {
case BIO_READ:
if (!(sc->sc_flags & G_ELI_FLAG_AUTH)) {
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
g_eli_crypto_read(sc, bp, 0);
break;
}
/* FALLTHROUGH */
case BIO_WRITE:
mtx_lock(&sc->sc_queue_mtx);
bioq_insert_tail(&sc->sc_queue, bp);
mtx_unlock(&sc->sc_queue_mtx);
wakeup(sc);
break;
case BIO_GETATTR:
case BIO_FLUSH:
case BIO_DELETE:
case BIO_SPEEDUP:
Add support for managing Shingled Magnetic Recording (SMR) drives. This change includes support for SCSI SMR drives (which conform to the Zoned Block Commands or ZBC spec) and ATA SMR drives (which conform to the Zoned ATA Command Set or ZAC spec) behind SAS expanders. This includes full management support through the GEOM BIO interface, and through a new userland utility, zonectl(8), and through camcontrol(8). This is now ready for filesystems to use to detect and manage zoned drives. (There is no work in progress that I know of to use this for ZFS or UFS, if anyone is interested, let me know and I may have some suggestions.) Also, improve ATA command passthrough and dispatch support, both via ATA and ATA passthrough over SCSI. Also, add support to camcontrol(8) for the ATA Extended Power Conditions feature set. You can now manage ATA device power states, and set various idle time thresholds for a drive to enter lower power states. Note that this change cannot be MFCed in full, because it depends on changes to the struct bio API that break compatilibity. In order to avoid breaking the stable API, only changes that don't touch or depend on the struct bio changes can be merged. For example, the camcontrol(8) changes don't depend on the new bio API, but zonectl(8) and the probe changes to the da(4) and ada(4) drivers do depend on it. Also note that the SMR changes have not yet been tested with an actual SCSI ZBC device, or a SCSI to ATA translation layer (SAT) that supports ZBC to ZAC translation. I have not yet gotten a suitable drive or SAT layer, so any testing help would be appreciated. These changes have been tested with Seagate Host Aware SATA drives attached to both SAS and SATA controllers. Also, I do not have any SATA Host Managed devices, and I suspect that it may take additional (hopefully minor) changes to support them. Thanks to Seagate for supplying the test hardware and answering questions. sbin/camcontrol/Makefile: Add epc.c and zone.c. sbin/camcontrol/camcontrol.8: Document the zone and epc subcommands. sbin/camcontrol/camcontrol.c: Add the zone and epc subcommands. Add auxiliary register support to build_ata_cmd(). Make sure to set the CAM_ATAIO_NEEDRESULT, CAM_ATAIO_DMA, and CAM_ATAIO_FPDMA flags as appropriate for ATA commands. Add a new get_ata_status() function to parse ATA result from SCSI sense descriptors (for ATA passthrough over SCSI) and ATA I/O requests. sbin/camcontrol/camcontrol.h: Update the build_ata_cmd() prototype Add get_ata_status(), zone(), and epc(). sbin/camcontrol/epc.c: Support for ATA Extended Power Conditions features. This includes support for all features documented in the ACS-4 Revision 12 specification from t13.org (dated February 18, 2016). The EPC feature set allows putting a drive into a power power mode immediately, or setting timeouts so that the drive will automatically enter progressively lower power states after various idle times. sbin/camcontrol/fwdownload.c: Update the firmware download code for the new build_ata_cmd() arguments. sbin/camcontrol/zone.c: Implement support for Shingled Magnetic Recording (SMR) drives via SCSI Zoned Block Commands (ZBC) and ATA Zoned Device ATA Command Set (ZAC). These specs were developed in concert, and are functionally identical. The primary differences are due to SCSI and ATA differences. (SCSI is big endian, ATA is little endian, for example.) This includes support for all commands defined in the ZBC and ZAC specs. sys/cam/ata/ata_all.c: Decode a number of additional ATA command names in ata_op_string(). Add a new CCB building function, ata_read_log(). Add ata_zac_mgmt_in() and ata_zac_mgmt_out() CCB building functions. These support both DMA and NCQ encapsulation. sys/cam/ata/ata_all.h: Add prototypes for ata_read_log(), ata_zac_mgmt_out(), and ata_zac_mgmt_in(). sys/cam/ata/ata_da.c: Revamp the ada(4) driver to support zoned devices. Add four new probe states to gather information needed for zone support. Add a new adasetflags() function to avoid duplication of large blocks of flag setting between the async handler and register functions. Add new sysctl variables that describe zone support and paramters. Add support for the new BIO_ZONE bio, and all of its subcommands: DISK_ZONE_OPEN, DISK_ZONE_CLOSE, DISK_ZONE_FINISH, DISK_ZONE_RWP, DISK_ZONE_REPORT_ZONES, and DISK_ZONE_GET_PARAMS. sys/cam/scsi/scsi_all.c: Add command descriptions for the ZBC IN/OUT commands. Add descriptions for ZBC Host Managed devices. Add a new function, scsi_ata_pass() to do ATA passthrough over SCSI. This will eventually replace scsi_ata_pass_16() -- it can create the 12, 16, and 32-byte variants of the ATA PASS-THROUGH command, and supports setting all of the registers defined as of SAT-4, Revision 5 (March 11, 2016). Change scsi_ata_identify() to use scsi_ata_pass() instead of scsi_ata_pass_16(). Add a new scsi_ata_read_log() function to facilitate reading ATA logs via SCSI. sys/cam/scsi/scsi_all.h: Add the new ATA PASS-THROUGH(32) command CDB. Add extended and variable CDB opcodes. Add Zoned Block Device Characteristics VPD page. Add ATA Return SCSI sense descriptor. Add prototypes for scsi_ata_read_log() and scsi_ata_pass(). sys/cam/scsi/scsi_da.c: Revamp the da(4) driver to support zoned devices. Add five new probe states, four of which are needed for ATA devices. Add five new sysctl variables that describe zone support and parameters. The da(4) driver supports SCSI ZBC devices, as well as ATA ZAC devices when they are attached via a SCSI to ATA Translation (SAT) layer. Since ZBC -> ZAC translation is a new feature in the T10 SAT-4 spec, most SATA drives will be supported via ATA commands sent via the SCSI ATA PASS-THROUGH command. The da(4) driver will prefer the ZBC interface, if it is available, for performance reasons, but will use the ATA PASS-THROUGH interface to the ZAC command set if the SAT layer doesn't support translation yet. As I mentioned above, ZBC command support is untested. Add support for the new BIO_ZONE bio, and all of its subcommands: DISK_ZONE_OPEN, DISK_ZONE_CLOSE, DISK_ZONE_FINISH, DISK_ZONE_RWP, DISK_ZONE_REPORT_ZONES, and DISK_ZONE_GET_PARAMS. Add scsi_zbc_in() and scsi_zbc_out() CCB building functions. Add scsi_ata_zac_mgmt_out() and scsi_ata_zac_mgmt_in() CCB/CDB building functions. Note that these have return values, unlike almost all other CCB building functions in CAM. The reason is that they can fail, depending upon the particular combination of input parameters. The primary failure case is if the user wants NCQ, but fails to specify additional CDB storage. NCQ requires using the 32-byte version of the SCSI ATA PASS-THROUGH command, and the current CAM CDB size is 16 bytes. sys/cam/scsi/scsi_da.h: Add ZBC IN and ZBC OUT CDBs and opcodes. Add SCSI Report Zones data structures. Add scsi_zbc_in(), scsi_zbc_out(), scsi_ata_zac_mgmt_out(), and scsi_ata_zac_mgmt_in() prototypes. sys/dev/ahci/ahci.c: Fix SEND / RECEIVE FPDMA QUEUED in the ahci(4) driver. ahci_setup_fis() previously set the top bits of the sector count register in the FIS to 0 for FPDMA commands. This is okay for read and write, because the PRIO field is in the only thing in those bits, and we don't implement that further up the stack. But, for SEND and RECEIVE FPDMA QUEUED, the subcommand is in that byte, so it needs to be transmitted to the drive. In ahci_setup_fis(), always set the the top 8 bits of the sector count register. We need it in both the standard and NCQ / FPDMA cases. sys/geom/eli/g_eli.c: Pass BIO_ZONE commands through the GELI class. sys/geom/geom.h: Add g_io_zonecmd() prototype. sys/geom/geom_dev.c: Add new DIOCZONECMD ioctl, which allows sending zone commands to disks. sys/geom/geom_disk.c: Add support for BIO_ZONE commands. sys/geom/geom_disk.h: Add a new flag, DISKFLAG_CANZONE, that indicates that a given GEOM disk client can handle BIO_ZONE commands. sys/geom/geom_io.c: Add a new function, g_io_zonecmd(), that handles execution of BIO_ZONE commands. Add permissions check for BIO_ZONE commands. Add command decoding for BIO_ZONE commands. sys/geom/geom_subr.c: Add DDB command decoding for BIO_ZONE commands. sys/kern/subr_devstat.c: Record statistics for REPORT ZONES commands. Note that the number of bytes transferred for REPORT ZONES won't quite match what is received from the harware. This is because we're necessarily counting bytes coming from the da(4) / ada(4) drivers, which are using the disk_zone.h interface to communicate up the stack. The structure sizes it uses are slightly different than the SCSI and ATA structure sizes. sys/sys/ata.h: Add many bit and structure definitions for ZAC, NCQ, and EPC command support. sys/sys/bio.h: Convert the bio_cmd field to a straight enumeration. This will yield more space for additional commands in the future. After change r297955 and other related changes, this is now possible. Converting to an enumeration will also prevent use as a bitmask in the future. sys/sys/disk.h: Define the DIOCZONECMD ioctl. sys/sys/disk_zone.h: Add a new API for managing zoned disks. This is very close to the SCSI ZBC and ATA ZAC standards, but uses integers in native byte order instead of big endian (SCSI) or little endian (ATA) byte arrays. This is intended to offer to the complete feature set of the ZBC and ZAC disk management without requiring the application developer to include SCSI or ATA headers. We also use one set of headers for ioctl consumers and kernel bio-level consumers. sys/sys/param.h: Bump __FreeBSD_version for sys/bio.h command changes, and inclusion of SMR support. usr.sbin/Makefile: Add the zonectl utility. usr.sbin/diskinfo/diskinfo.c Add disk zoning capability to the 'diskinfo -v' output. usr.sbin/zonectl/Makefile: Add zonectl makefile. usr.sbin/zonectl/zonectl.8 zonectl(8) man page. usr.sbin/zonectl/zonectl.c The zonectl(8) utility. This allows managing SCSI or ATA zoned disks via the disk_zone.h API. You can report zones, reset write pointers, get parameters, etc. Sponsored by: Spectra Logic Differential Revision: https://reviews.freebsd.org/D6147 Reviewed by: wblock (documentation)
2016-05-19 14:08:36 +00:00
case BIO_ZONE:
if (bp->bio_cmd == BIO_GETATTR)
cbp->bio_done = g_eli_getattr_done;
else
cbp->bio_done = g_std_done;
cp = LIST_FIRST(&sc->sc_geom->consumer);
cbp->bio_to = cp->provider;
G_ELI_LOGREQ(2, cbp, "Sending request.");
g_io_request(cbp, cp);
break;
}
}
static int
g_eli_newsession(struct g_eli_worker *wr)
{
struct g_eli_softc *sc;
Refactor driver and consumer interfaces for OCF (in-kernel crypto). - The linked list of cryptoini structures used in session initialization is replaced with a new flat structure: struct crypto_session_params. This session includes a new mode to define how the other fields should be interpreted. Available modes include: - COMPRESS (for compression/decompression) - CIPHER (for simply encryption/decryption) - DIGEST (computing and verifying digests) - AEAD (combined auth and encryption such as AES-GCM and AES-CCM) - ETA (combined auth and encryption using encrypt-then-authenticate) Additional modes could be added in the future (e.g. if we wanted to support TLS MtE for AES-CBC in the kernel we could add a new mode for that. TLS modes might also affect how AAD is interpreted, etc.) The flat structure also includes the key lengths and algorithms as before. However, code doesn't have to walk the linked list and switch on the algorithm to determine which key is the auth key vs encryption key. The 'csp_auth_*' fields are always used for auth keys and settings and 'csp_cipher_*' for cipher. (Compression algorithms are stored in csp_cipher_alg.) - Drivers no longer register a list of supported algorithms. This doesn't quite work when you factor in modes (e.g. a driver might support both AES-CBC and SHA2-256-HMAC separately but not combined for ETA). Instead, a new 'crypto_probesession' method has been added to the kobj interface for symmteric crypto drivers. This method returns a negative value on success (similar to how device_probe works) and the crypto framework uses this value to pick the "best" driver. There are three constants for hardware (e.g. ccr), accelerated software (e.g. aesni), and plain software (cryptosoft) that give preference in that order. One effect of this is that if you request only hardware when creating a new session, you will no longer get a session using accelerated software. Another effect is that the default setting to disallow software crypto via /dev/crypto now disables accelerated software. Once a driver is chosen, 'crypto_newsession' is invoked as before. - Crypto operations are now solely described by the flat 'cryptop' structure. The linked list of descriptors has been removed. A separate enum has been added to describe the type of data buffer in use instead of using CRYPTO_F_* flags to make it easier to add more types in the future if needed (e.g. wired userspace buffers for zero-copy). It will also make it easier to re-introduce separate input and output buffers (in-kernel TLS would benefit from this). Try to make the flags related to IV handling less insane: - CRYPTO_F_IV_SEPARATE means that the IV is stored in the 'crp_iv' member of the operation structure. If this flag is not set, the IV is stored in the data buffer at the 'crp_iv_start' offset. - CRYPTO_F_IV_GENERATE means that a random IV should be generated and stored into the data buffer. This cannot be used with CRYPTO_F_IV_SEPARATE. If a consumer wants to deal with explicit vs implicit IVs, etc. it can always generate the IV however it needs and store partial IVs in the buffer and the full IV/nonce in crp_iv and set CRYPTO_F_IV_SEPARATE. The layout of the buffer is now described via fields in cryptop. crp_aad_start and crp_aad_length define the boundaries of any AAD. Previously with GCM and CCM you defined an auth crd with this range, but for ETA your auth crd had to span both the AAD and plaintext (and they had to be adjacent). crp_payload_start and crp_payload_length define the boundaries of the plaintext/ciphertext. Modes that only do a single operation (COMPRESS, CIPHER, DIGEST) should only use this region and leave the AAD region empty. If a digest is present (or should be generated), it's starting location is marked by crp_digest_start. Instead of using the CRD_F_ENCRYPT flag to determine the direction of the operation, cryptop now includes an 'op' field defining the operation to perform. For digests I've added a new VERIFY digest mode which assumes a digest is present in the input and fails the request with EBADMSG if it doesn't match the internally-computed digest. GCM and CCM already assumed this, and the new AEAD mode requires this for decryption. The new ETA mode now also requires this for decryption, so IPsec and GELI no longer do their own authentication verification. Simple DIGEST operations can also do this, though there are no in-tree consumers. To eventually support some refcounting to close races, the session cookie is now passed to crypto_getop() and clients should no longer set crp_sesssion directly. - Assymteric crypto operation structures should be allocated via crypto_getkreq() and freed via crypto_freekreq(). This permits the crypto layer to track open asym requests and close races with a driver trying to unregister while asym requests are in flight. - crypto_copyback, crypto_copydata, crypto_apply, and crypto_contiguous_subsegment now accept the 'crp' object as the first parameter instead of individual members. This makes it easier to deal with different buffer types in the future as well as separate input and output buffers. It's also simpler for driver writers to use. - bus_dmamap_load_crp() loads a DMA mapping for a crypto buffer. This understands the various types of buffers so that drivers that use DMA do not have to be aware of different buffer types. - Helper routines now exist to build an auth context for HMAC IPAD and OPAD. This reduces some duplicated work among drivers. - Key buffers are now treated as const throughout the framework and in device drivers. However, session key buffers provided when a session is created are expected to remain alive for the duration of the session. - GCM and CCM sessions now only specify a cipher algorithm and a cipher key. The redundant auth information is not needed or used. - For cryptosoft, split up the code a bit such that the 'process' callback now invokes a function pointer in the session. This function pointer is set based on the mode (in effect) though it simplifies a few edge cases that would otherwise be in the switch in 'process'. It does split up GCM vs CCM which I think is more readable even if there is some duplication. - I changed /dev/crypto to support GMAC requests using CRYPTO_AES_NIST_GMAC as an auth algorithm and updated cryptocheck to work with it. - Combined cipher and auth sessions via /dev/crypto now always use ETA mode. The COP_F_CIPHER_FIRST flag is now a no-op that is ignored. This was actually documented as being true in crypto(4) before, but the code had not implemented this before I added the CIPHER_FIRST flag. - I have not yet updated /dev/crypto to be aware of explicit modes for sessions. I will probably do that at some point in the future as well as teach it about IV/nonce and tag lengths for AEAD so we can support all of the NIST KAT tests for GCM and CCM. - I've split up the exising crypto.9 manpage into several pages of which many are written from scratch. - I have converted all drivers and consumers in the tree and verified that they compile, but I have not tested all of them. I have tested the following drivers: - cryptosoft - aesni (AES only) - blake2 - ccr and the following consumers: - cryptodev - IPsec - ktls_ocf - GELI (lightly) I have not tested the following: - ccp - aesni with sha - hifn - kgssapi_krb5 - ubsec - padlock - safe - armv8_crypto (aarch64) - glxsb (i386) - sec (ppc) - cesa (armv7) - cryptocteon (mips64) - nlmsec (mips64) Discussed with: cem Relnotes: yes Sponsored by: Chelsio Communications Differential Revision: https://reviews.freebsd.org/D23677
2020-03-27 18:25:23 +00:00
struct crypto_session_params csp;
uint32_t caps;
int error, new_crypto;
Refactor driver and consumer interfaces for OCF (in-kernel crypto). - The linked list of cryptoini structures used in session initialization is replaced with a new flat structure: struct crypto_session_params. This session includes a new mode to define how the other fields should be interpreted. Available modes include: - COMPRESS (for compression/decompression) - CIPHER (for simply encryption/decryption) - DIGEST (computing and verifying digests) - AEAD (combined auth and encryption such as AES-GCM and AES-CCM) - ETA (combined auth and encryption using encrypt-then-authenticate) Additional modes could be added in the future (e.g. if we wanted to support TLS MtE for AES-CBC in the kernel we could add a new mode for that. TLS modes might also affect how AAD is interpreted, etc.) The flat structure also includes the key lengths and algorithms as before. However, code doesn't have to walk the linked list and switch on the algorithm to determine which key is the auth key vs encryption key. The 'csp_auth_*' fields are always used for auth keys and settings and 'csp_cipher_*' for cipher. (Compression algorithms are stored in csp_cipher_alg.) - Drivers no longer register a list of supported algorithms. This doesn't quite work when you factor in modes (e.g. a driver might support both AES-CBC and SHA2-256-HMAC separately but not combined for ETA). Instead, a new 'crypto_probesession' method has been added to the kobj interface for symmteric crypto drivers. This method returns a negative value on success (similar to how device_probe works) and the crypto framework uses this value to pick the "best" driver. There are three constants for hardware (e.g. ccr), accelerated software (e.g. aesni), and plain software (cryptosoft) that give preference in that order. One effect of this is that if you request only hardware when creating a new session, you will no longer get a session using accelerated software. Another effect is that the default setting to disallow software crypto via /dev/crypto now disables accelerated software. Once a driver is chosen, 'crypto_newsession' is invoked as before. - Crypto operations are now solely described by the flat 'cryptop' structure. The linked list of descriptors has been removed. A separate enum has been added to describe the type of data buffer in use instead of using CRYPTO_F_* flags to make it easier to add more types in the future if needed (e.g. wired userspace buffers for zero-copy). It will also make it easier to re-introduce separate input and output buffers (in-kernel TLS would benefit from this). Try to make the flags related to IV handling less insane: - CRYPTO_F_IV_SEPARATE means that the IV is stored in the 'crp_iv' member of the operation structure. If this flag is not set, the IV is stored in the data buffer at the 'crp_iv_start' offset. - CRYPTO_F_IV_GENERATE means that a random IV should be generated and stored into the data buffer. This cannot be used with CRYPTO_F_IV_SEPARATE. If a consumer wants to deal with explicit vs implicit IVs, etc. it can always generate the IV however it needs and store partial IVs in the buffer and the full IV/nonce in crp_iv and set CRYPTO_F_IV_SEPARATE. The layout of the buffer is now described via fields in cryptop. crp_aad_start and crp_aad_length define the boundaries of any AAD. Previously with GCM and CCM you defined an auth crd with this range, but for ETA your auth crd had to span both the AAD and plaintext (and they had to be adjacent). crp_payload_start and crp_payload_length define the boundaries of the plaintext/ciphertext. Modes that only do a single operation (COMPRESS, CIPHER, DIGEST) should only use this region and leave the AAD region empty. If a digest is present (or should be generated), it's starting location is marked by crp_digest_start. Instead of using the CRD_F_ENCRYPT flag to determine the direction of the operation, cryptop now includes an 'op' field defining the operation to perform. For digests I've added a new VERIFY digest mode which assumes a digest is present in the input and fails the request with EBADMSG if it doesn't match the internally-computed digest. GCM and CCM already assumed this, and the new AEAD mode requires this for decryption. The new ETA mode now also requires this for decryption, so IPsec and GELI no longer do their own authentication verification. Simple DIGEST operations can also do this, though there are no in-tree consumers. To eventually support some refcounting to close races, the session cookie is now passed to crypto_getop() and clients should no longer set crp_sesssion directly. - Assymteric crypto operation structures should be allocated via crypto_getkreq() and freed via crypto_freekreq(). This permits the crypto layer to track open asym requests and close races with a driver trying to unregister while asym requests are in flight. - crypto_copyback, crypto_copydata, crypto_apply, and crypto_contiguous_subsegment now accept the 'crp' object as the first parameter instead of individual members. This makes it easier to deal with different buffer types in the future as well as separate input and output buffers. It's also simpler for driver writers to use. - bus_dmamap_load_crp() loads a DMA mapping for a crypto buffer. This understands the various types of buffers so that drivers that use DMA do not have to be aware of different buffer types. - Helper routines now exist to build an auth context for HMAC IPAD and OPAD. This reduces some duplicated work among drivers. - Key buffers are now treated as const throughout the framework and in device drivers. However, session key buffers provided when a session is created are expected to remain alive for the duration of the session. - GCM and CCM sessions now only specify a cipher algorithm and a cipher key. The redundant auth information is not needed or used. - For cryptosoft, split up the code a bit such that the 'process' callback now invokes a function pointer in the session. This function pointer is set based on the mode (in effect) though it simplifies a few edge cases that would otherwise be in the switch in 'process'. It does split up GCM vs CCM which I think is more readable even if there is some duplication. - I changed /dev/crypto to support GMAC requests using CRYPTO_AES_NIST_GMAC as an auth algorithm and updated cryptocheck to work with it. - Combined cipher and auth sessions via /dev/crypto now always use ETA mode. The COP_F_CIPHER_FIRST flag is now a no-op that is ignored. This was actually documented as being true in crypto(4) before, but the code had not implemented this before I added the CIPHER_FIRST flag. - I have not yet updated /dev/crypto to be aware of explicit modes for sessions. I will probably do that at some point in the future as well as teach it about IV/nonce and tag lengths for AEAD so we can support all of the NIST KAT tests for GCM and CCM. - I've split up the exising crypto.9 manpage into several pages of which many are written from scratch. - I have converted all drivers and consumers in the tree and verified that they compile, but I have not tested all of them. I have tested the following drivers: - cryptosoft - aesni (AES only) - blake2 - ccr and the following consumers: - cryptodev - IPsec - ktls_ocf - GELI (lightly) I have not tested the following: - ccp - aesni with sha - hifn - kgssapi_krb5 - ubsec - padlock - safe - armv8_crypto (aarch64) - glxsb (i386) - sec (ppc) - cesa (armv7) - cryptocteon (mips64) - nlmsec (mips64) Discussed with: cem Relnotes: yes Sponsored by: Chelsio Communications Differential Revision: https://reviews.freebsd.org/D23677
2020-03-27 18:25:23 +00:00
void *key;
sc = wr->w_softc;
Refactor driver and consumer interfaces for OCF (in-kernel crypto). - The linked list of cryptoini structures used in session initialization is replaced with a new flat structure: struct crypto_session_params. This session includes a new mode to define how the other fields should be interpreted. Available modes include: - COMPRESS (for compression/decompression) - CIPHER (for simply encryption/decryption) - DIGEST (computing and verifying digests) - AEAD (combined auth and encryption such as AES-GCM and AES-CCM) - ETA (combined auth and encryption using encrypt-then-authenticate) Additional modes could be added in the future (e.g. if we wanted to support TLS MtE for AES-CBC in the kernel we could add a new mode for that. TLS modes might also affect how AAD is interpreted, etc.) The flat structure also includes the key lengths and algorithms as before. However, code doesn't have to walk the linked list and switch on the algorithm to determine which key is the auth key vs encryption key. The 'csp_auth_*' fields are always used for auth keys and settings and 'csp_cipher_*' for cipher. (Compression algorithms are stored in csp_cipher_alg.) - Drivers no longer register a list of supported algorithms. This doesn't quite work when you factor in modes (e.g. a driver might support both AES-CBC and SHA2-256-HMAC separately but not combined for ETA). Instead, a new 'crypto_probesession' method has been added to the kobj interface for symmteric crypto drivers. This method returns a negative value on success (similar to how device_probe works) and the crypto framework uses this value to pick the "best" driver. There are three constants for hardware (e.g. ccr), accelerated software (e.g. aesni), and plain software (cryptosoft) that give preference in that order. One effect of this is that if you request only hardware when creating a new session, you will no longer get a session using accelerated software. Another effect is that the default setting to disallow software crypto via /dev/crypto now disables accelerated software. Once a driver is chosen, 'crypto_newsession' is invoked as before. - Crypto operations are now solely described by the flat 'cryptop' structure. The linked list of descriptors has been removed. A separate enum has been added to describe the type of data buffer in use instead of using CRYPTO_F_* flags to make it easier to add more types in the future if needed (e.g. wired userspace buffers for zero-copy). It will also make it easier to re-introduce separate input and output buffers (in-kernel TLS would benefit from this). Try to make the flags related to IV handling less insane: - CRYPTO_F_IV_SEPARATE means that the IV is stored in the 'crp_iv' member of the operation structure. If this flag is not set, the IV is stored in the data buffer at the 'crp_iv_start' offset. - CRYPTO_F_IV_GENERATE means that a random IV should be generated and stored into the data buffer. This cannot be used with CRYPTO_F_IV_SEPARATE. If a consumer wants to deal with explicit vs implicit IVs, etc. it can always generate the IV however it needs and store partial IVs in the buffer and the full IV/nonce in crp_iv and set CRYPTO_F_IV_SEPARATE. The layout of the buffer is now described via fields in cryptop. crp_aad_start and crp_aad_length define the boundaries of any AAD. Previously with GCM and CCM you defined an auth crd with this range, but for ETA your auth crd had to span both the AAD and plaintext (and they had to be adjacent). crp_payload_start and crp_payload_length define the boundaries of the plaintext/ciphertext. Modes that only do a single operation (COMPRESS, CIPHER, DIGEST) should only use this region and leave the AAD region empty. If a digest is present (or should be generated), it's starting location is marked by crp_digest_start. Instead of using the CRD_F_ENCRYPT flag to determine the direction of the operation, cryptop now includes an 'op' field defining the operation to perform. For digests I've added a new VERIFY digest mode which assumes a digest is present in the input and fails the request with EBADMSG if it doesn't match the internally-computed digest. GCM and CCM already assumed this, and the new AEAD mode requires this for decryption. The new ETA mode now also requires this for decryption, so IPsec and GELI no longer do their own authentication verification. Simple DIGEST operations can also do this, though there are no in-tree consumers. To eventually support some refcounting to close races, the session cookie is now passed to crypto_getop() and clients should no longer set crp_sesssion directly. - Assymteric crypto operation structures should be allocated via crypto_getkreq() and freed via crypto_freekreq(). This permits the crypto layer to track open asym requests and close races with a driver trying to unregister while asym requests are in flight. - crypto_copyback, crypto_copydata, crypto_apply, and crypto_contiguous_subsegment now accept the 'crp' object as the first parameter instead of individual members. This makes it easier to deal with different buffer types in the future as well as separate input and output buffers. It's also simpler for driver writers to use. - bus_dmamap_load_crp() loads a DMA mapping for a crypto buffer. This understands the various types of buffers so that drivers that use DMA do not have to be aware of different buffer types. - Helper routines now exist to build an auth context for HMAC IPAD and OPAD. This reduces some duplicated work among drivers. - Key buffers are now treated as const throughout the framework and in device drivers. However, session key buffers provided when a session is created are expected to remain alive for the duration of the session. - GCM and CCM sessions now only specify a cipher algorithm and a cipher key. The redundant auth information is not needed or used. - For cryptosoft, split up the code a bit such that the 'process' callback now invokes a function pointer in the session. This function pointer is set based on the mode (in effect) though it simplifies a few edge cases that would otherwise be in the switch in 'process'. It does split up GCM vs CCM which I think is more readable even if there is some duplication. - I changed /dev/crypto to support GMAC requests using CRYPTO_AES_NIST_GMAC as an auth algorithm and updated cryptocheck to work with it. - Combined cipher and auth sessions via /dev/crypto now always use ETA mode. The COP_F_CIPHER_FIRST flag is now a no-op that is ignored. This was actually documented as being true in crypto(4) before, but the code had not implemented this before I added the CIPHER_FIRST flag. - I have not yet updated /dev/crypto to be aware of explicit modes for sessions. I will probably do that at some point in the future as well as teach it about IV/nonce and tag lengths for AEAD so we can support all of the NIST KAT tests for GCM and CCM. - I've split up the exising crypto.9 manpage into several pages of which many are written from scratch. - I have converted all drivers and consumers in the tree and verified that they compile, but I have not tested all of them. I have tested the following drivers: - cryptosoft - aesni (AES only) - blake2 - ccr and the following consumers: - cryptodev - IPsec - ktls_ocf - GELI (lightly) I have not tested the following: - ccp - aesni with sha - hifn - kgssapi_krb5 - ubsec - padlock - safe - armv8_crypto (aarch64) - glxsb (i386) - sec (ppc) - cesa (armv7) - cryptocteon (mips64) - nlmsec (mips64) Discussed with: cem Relnotes: yes Sponsored by: Chelsio Communications Differential Revision: https://reviews.freebsd.org/D23677
2020-03-27 18:25:23 +00:00
memset(&csp, 0, sizeof(csp));
csp.csp_mode = CSP_MODE_CIPHER;
csp.csp_cipher_alg = sc->sc_ealgo;
csp.csp_ivlen = g_eli_ivlen(sc->sc_ealgo);
csp.csp_cipher_klen = sc->sc_ekeylen / 8;
if (sc->sc_ealgo == CRYPTO_AES_XTS)
Refactor driver and consumer interfaces for OCF (in-kernel crypto). - The linked list of cryptoini structures used in session initialization is replaced with a new flat structure: struct crypto_session_params. This session includes a new mode to define how the other fields should be interpreted. Available modes include: - COMPRESS (for compression/decompression) - CIPHER (for simply encryption/decryption) - DIGEST (computing and verifying digests) - AEAD (combined auth and encryption such as AES-GCM and AES-CCM) - ETA (combined auth and encryption using encrypt-then-authenticate) Additional modes could be added in the future (e.g. if we wanted to support TLS MtE for AES-CBC in the kernel we could add a new mode for that. TLS modes might also affect how AAD is interpreted, etc.) The flat structure also includes the key lengths and algorithms as before. However, code doesn't have to walk the linked list and switch on the algorithm to determine which key is the auth key vs encryption key. The 'csp_auth_*' fields are always used for auth keys and settings and 'csp_cipher_*' for cipher. (Compression algorithms are stored in csp_cipher_alg.) - Drivers no longer register a list of supported algorithms. This doesn't quite work when you factor in modes (e.g. a driver might support both AES-CBC and SHA2-256-HMAC separately but not combined for ETA). Instead, a new 'crypto_probesession' method has been added to the kobj interface for symmteric crypto drivers. This method returns a negative value on success (similar to how device_probe works) and the crypto framework uses this value to pick the "best" driver. There are three constants for hardware (e.g. ccr), accelerated software (e.g. aesni), and plain software (cryptosoft) that give preference in that order. One effect of this is that if you request only hardware when creating a new session, you will no longer get a session using accelerated software. Another effect is that the default setting to disallow software crypto via /dev/crypto now disables accelerated software. Once a driver is chosen, 'crypto_newsession' is invoked as before. - Crypto operations are now solely described by the flat 'cryptop' structure. The linked list of descriptors has been removed. A separate enum has been added to describe the type of data buffer in use instead of using CRYPTO_F_* flags to make it easier to add more types in the future if needed (e.g. wired userspace buffers for zero-copy). It will also make it easier to re-introduce separate input and output buffers (in-kernel TLS would benefit from this). Try to make the flags related to IV handling less insane: - CRYPTO_F_IV_SEPARATE means that the IV is stored in the 'crp_iv' member of the operation structure. If this flag is not set, the IV is stored in the data buffer at the 'crp_iv_start' offset. - CRYPTO_F_IV_GENERATE means that a random IV should be generated and stored into the data buffer. This cannot be used with CRYPTO_F_IV_SEPARATE. If a consumer wants to deal with explicit vs implicit IVs, etc. it can always generate the IV however it needs and store partial IVs in the buffer and the full IV/nonce in crp_iv and set CRYPTO_F_IV_SEPARATE. The layout of the buffer is now described via fields in cryptop. crp_aad_start and crp_aad_length define the boundaries of any AAD. Previously with GCM and CCM you defined an auth crd with this range, but for ETA your auth crd had to span both the AAD and plaintext (and they had to be adjacent). crp_payload_start and crp_payload_length define the boundaries of the plaintext/ciphertext. Modes that only do a single operation (COMPRESS, CIPHER, DIGEST) should only use this region and leave the AAD region empty. If a digest is present (or should be generated), it's starting location is marked by crp_digest_start. Instead of using the CRD_F_ENCRYPT flag to determine the direction of the operation, cryptop now includes an 'op' field defining the operation to perform. For digests I've added a new VERIFY digest mode which assumes a digest is present in the input and fails the request with EBADMSG if it doesn't match the internally-computed digest. GCM and CCM already assumed this, and the new AEAD mode requires this for decryption. The new ETA mode now also requires this for decryption, so IPsec and GELI no longer do their own authentication verification. Simple DIGEST operations can also do this, though there are no in-tree consumers. To eventually support some refcounting to close races, the session cookie is now passed to crypto_getop() and clients should no longer set crp_sesssion directly. - Assymteric crypto operation structures should be allocated via crypto_getkreq() and freed via crypto_freekreq(). This permits the crypto layer to track open asym requests and close races with a driver trying to unregister while asym requests are in flight. - crypto_copyback, crypto_copydata, crypto_apply, and crypto_contiguous_subsegment now accept the 'crp' object as the first parameter instead of individual members. This makes it easier to deal with different buffer types in the future as well as separate input and output buffers. It's also simpler for driver writers to use. - bus_dmamap_load_crp() loads a DMA mapping for a crypto buffer. This understands the various types of buffers so that drivers that use DMA do not have to be aware of different buffer types. - Helper routines now exist to build an auth context for HMAC IPAD and OPAD. This reduces some duplicated work among drivers. - Key buffers are now treated as const throughout the framework and in device drivers. However, session key buffers provided when a session is created are expected to remain alive for the duration of the session. - GCM and CCM sessions now only specify a cipher algorithm and a cipher key. The redundant auth information is not needed or used. - For cryptosoft, split up the code a bit such that the 'process' callback now invokes a function pointer in the session. This function pointer is set based on the mode (in effect) though it simplifies a few edge cases that would otherwise be in the switch in 'process'. It does split up GCM vs CCM which I think is more readable even if there is some duplication. - I changed /dev/crypto to support GMAC requests using CRYPTO_AES_NIST_GMAC as an auth algorithm and updated cryptocheck to work with it. - Combined cipher and auth sessions via /dev/crypto now always use ETA mode. The COP_F_CIPHER_FIRST flag is now a no-op that is ignored. This was actually documented as being true in crypto(4) before, but the code had not implemented this before I added the CIPHER_FIRST flag. - I have not yet updated /dev/crypto to be aware of explicit modes for sessions. I will probably do that at some point in the future as well as teach it about IV/nonce and tag lengths for AEAD so we can support all of the NIST KAT tests for GCM and CCM. - I've split up the exising crypto.9 manpage into several pages of which many are written from scratch. - I have converted all drivers and consumers in the tree and verified that they compile, but I have not tested all of them. I have tested the following drivers: - cryptosoft - aesni (AES only) - blake2 - ccr and the following consumers: - cryptodev - IPsec - ktls_ocf - GELI (lightly) I have not tested the following: - ccp - aesni with sha - hifn - kgssapi_krb5 - ubsec - padlock - safe - armv8_crypto (aarch64) - glxsb (i386) - sec (ppc) - cesa (armv7) - cryptocteon (mips64) - nlmsec (mips64) Discussed with: cem Relnotes: yes Sponsored by: Chelsio Communications Differential Revision: https://reviews.freebsd.org/D23677
2020-03-27 18:25:23 +00:00
csp.csp_cipher_klen <<= 1;
if ((sc->sc_flags & G_ELI_FLAG_FIRST_KEY) != 0) {
Refactor driver and consumer interfaces for OCF (in-kernel crypto). - The linked list of cryptoini structures used in session initialization is replaced with a new flat structure: struct crypto_session_params. This session includes a new mode to define how the other fields should be interpreted. Available modes include: - COMPRESS (for compression/decompression) - CIPHER (for simply encryption/decryption) - DIGEST (computing and verifying digests) - AEAD (combined auth and encryption such as AES-GCM and AES-CCM) - ETA (combined auth and encryption using encrypt-then-authenticate) Additional modes could be added in the future (e.g. if we wanted to support TLS MtE for AES-CBC in the kernel we could add a new mode for that. TLS modes might also affect how AAD is interpreted, etc.) The flat structure also includes the key lengths and algorithms as before. However, code doesn't have to walk the linked list and switch on the algorithm to determine which key is the auth key vs encryption key. The 'csp_auth_*' fields are always used for auth keys and settings and 'csp_cipher_*' for cipher. (Compression algorithms are stored in csp_cipher_alg.) - Drivers no longer register a list of supported algorithms. This doesn't quite work when you factor in modes (e.g. a driver might support both AES-CBC and SHA2-256-HMAC separately but not combined for ETA). Instead, a new 'crypto_probesession' method has been added to the kobj interface for symmteric crypto drivers. This method returns a negative value on success (similar to how device_probe works) and the crypto framework uses this value to pick the "best" driver. There are three constants for hardware (e.g. ccr), accelerated software (e.g. aesni), and plain software (cryptosoft) that give preference in that order. One effect of this is that if you request only hardware when creating a new session, you will no longer get a session using accelerated software. Another effect is that the default setting to disallow software crypto via /dev/crypto now disables accelerated software. Once a driver is chosen, 'crypto_newsession' is invoked as before. - Crypto operations are now solely described by the flat 'cryptop' structure. The linked list of descriptors has been removed. A separate enum has been added to describe the type of data buffer in use instead of using CRYPTO_F_* flags to make it easier to add more types in the future if needed (e.g. wired userspace buffers for zero-copy). It will also make it easier to re-introduce separate input and output buffers (in-kernel TLS would benefit from this). Try to make the flags related to IV handling less insane: - CRYPTO_F_IV_SEPARATE means that the IV is stored in the 'crp_iv' member of the operation structure. If this flag is not set, the IV is stored in the data buffer at the 'crp_iv_start' offset. - CRYPTO_F_IV_GENERATE means that a random IV should be generated and stored into the data buffer. This cannot be used with CRYPTO_F_IV_SEPARATE. If a consumer wants to deal with explicit vs implicit IVs, etc. it can always generate the IV however it needs and store partial IVs in the buffer and the full IV/nonce in crp_iv and set CRYPTO_F_IV_SEPARATE. The layout of the buffer is now described via fields in cryptop. crp_aad_start and crp_aad_length define the boundaries of any AAD. Previously with GCM and CCM you defined an auth crd with this range, but for ETA your auth crd had to span both the AAD and plaintext (and they had to be adjacent). crp_payload_start and crp_payload_length define the boundaries of the plaintext/ciphertext. Modes that only do a single operation (COMPRESS, CIPHER, DIGEST) should only use this region and leave the AAD region empty. If a digest is present (or should be generated), it's starting location is marked by crp_digest_start. Instead of using the CRD_F_ENCRYPT flag to determine the direction of the operation, cryptop now includes an 'op' field defining the operation to perform. For digests I've added a new VERIFY digest mode which assumes a digest is present in the input and fails the request with EBADMSG if it doesn't match the internally-computed digest. GCM and CCM already assumed this, and the new AEAD mode requires this for decryption. The new ETA mode now also requires this for decryption, so IPsec and GELI no longer do their own authentication verification. Simple DIGEST operations can also do this, though there are no in-tree consumers. To eventually support some refcounting to close races, the session cookie is now passed to crypto_getop() and clients should no longer set crp_sesssion directly. - Assymteric crypto operation structures should be allocated via crypto_getkreq() and freed via crypto_freekreq(). This permits the crypto layer to track open asym requests and close races with a driver trying to unregister while asym requests are in flight. - crypto_copyback, crypto_copydata, crypto_apply, and crypto_contiguous_subsegment now accept the 'crp' object as the first parameter instead of individual members. This makes it easier to deal with different buffer types in the future as well as separate input and output buffers. It's also simpler for driver writers to use. - bus_dmamap_load_crp() loads a DMA mapping for a crypto buffer. This understands the various types of buffers so that drivers that use DMA do not have to be aware of different buffer types. - Helper routines now exist to build an auth context for HMAC IPAD and OPAD. This reduces some duplicated work among drivers. - Key buffers are now treated as const throughout the framework and in device drivers. However, session key buffers provided when a session is created are expected to remain alive for the duration of the session. - GCM and CCM sessions now only specify a cipher algorithm and a cipher key. The redundant auth information is not needed or used. - For cryptosoft, split up the code a bit such that the 'process' callback now invokes a function pointer in the session. This function pointer is set based on the mode (in effect) though it simplifies a few edge cases that would otherwise be in the switch in 'process'. It does split up GCM vs CCM which I think is more readable even if there is some duplication. - I changed /dev/crypto to support GMAC requests using CRYPTO_AES_NIST_GMAC as an auth algorithm and updated cryptocheck to work with it. - Combined cipher and auth sessions via /dev/crypto now always use ETA mode. The COP_F_CIPHER_FIRST flag is now a no-op that is ignored. This was actually documented as being true in crypto(4) before, but the code had not implemented this before I added the CIPHER_FIRST flag. - I have not yet updated /dev/crypto to be aware of explicit modes for sessions. I will probably do that at some point in the future as well as teach it about IV/nonce and tag lengths for AEAD so we can support all of the NIST KAT tests for GCM and CCM. - I've split up the exising crypto.9 manpage into several pages of which many are written from scratch. - I have converted all drivers and consumers in the tree and verified that they compile, but I have not tested all of them. I have tested the following drivers: - cryptosoft - aesni (AES only) - blake2 - ccr and the following consumers: - cryptodev - IPsec - ktls_ocf - GELI (lightly) I have not tested the following: - ccp - aesni with sha - hifn - kgssapi_krb5 - ubsec - padlock - safe - armv8_crypto (aarch64) - glxsb (i386) - sec (ppc) - cesa (armv7) - cryptocteon (mips64) - nlmsec (mips64) Discussed with: cem Relnotes: yes Sponsored by: Chelsio Communications Differential Revision: https://reviews.freebsd.org/D23677
2020-03-27 18:25:23 +00:00
key = g_eli_key_hold(sc, 0,
LIST_FIRST(&sc->sc_geom->consumer)->provider->sectorsize);
Refactor driver and consumer interfaces for OCF (in-kernel crypto). - The linked list of cryptoini structures used in session initialization is replaced with a new flat structure: struct crypto_session_params. This session includes a new mode to define how the other fields should be interpreted. Available modes include: - COMPRESS (for compression/decompression) - CIPHER (for simply encryption/decryption) - DIGEST (computing and verifying digests) - AEAD (combined auth and encryption such as AES-GCM and AES-CCM) - ETA (combined auth and encryption using encrypt-then-authenticate) Additional modes could be added in the future (e.g. if we wanted to support TLS MtE for AES-CBC in the kernel we could add a new mode for that. TLS modes might also affect how AAD is interpreted, etc.) The flat structure also includes the key lengths and algorithms as before. However, code doesn't have to walk the linked list and switch on the algorithm to determine which key is the auth key vs encryption key. The 'csp_auth_*' fields are always used for auth keys and settings and 'csp_cipher_*' for cipher. (Compression algorithms are stored in csp_cipher_alg.) - Drivers no longer register a list of supported algorithms. This doesn't quite work when you factor in modes (e.g. a driver might support both AES-CBC and SHA2-256-HMAC separately but not combined for ETA). Instead, a new 'crypto_probesession' method has been added to the kobj interface for symmteric crypto drivers. This method returns a negative value on success (similar to how device_probe works) and the crypto framework uses this value to pick the "best" driver. There are three constants for hardware (e.g. ccr), accelerated software (e.g. aesni), and plain software (cryptosoft) that give preference in that order. One effect of this is that if you request only hardware when creating a new session, you will no longer get a session using accelerated software. Another effect is that the default setting to disallow software crypto via /dev/crypto now disables accelerated software. Once a driver is chosen, 'crypto_newsession' is invoked as before. - Crypto operations are now solely described by the flat 'cryptop' structure. The linked list of descriptors has been removed. A separate enum has been added to describe the type of data buffer in use instead of using CRYPTO_F_* flags to make it easier to add more types in the future if needed (e.g. wired userspace buffers for zero-copy). It will also make it easier to re-introduce separate input and output buffers (in-kernel TLS would benefit from this). Try to make the flags related to IV handling less insane: - CRYPTO_F_IV_SEPARATE means that the IV is stored in the 'crp_iv' member of the operation structure. If this flag is not set, the IV is stored in the data buffer at the 'crp_iv_start' offset. - CRYPTO_F_IV_GENERATE means that a random IV should be generated and stored into the data buffer. This cannot be used with CRYPTO_F_IV_SEPARATE. If a consumer wants to deal with explicit vs implicit IVs, etc. it can always generate the IV however it needs and store partial IVs in the buffer and the full IV/nonce in crp_iv and set CRYPTO_F_IV_SEPARATE. The layout of the buffer is now described via fields in cryptop. crp_aad_start and crp_aad_length define the boundaries of any AAD. Previously with GCM and CCM you defined an auth crd with this range, but for ETA your auth crd had to span both the AAD and plaintext (and they had to be adjacent). crp_payload_start and crp_payload_length define the boundaries of the plaintext/ciphertext. Modes that only do a single operation (COMPRESS, CIPHER, DIGEST) should only use this region and leave the AAD region empty. If a digest is present (or should be generated), it's starting location is marked by crp_digest_start. Instead of using the CRD_F_ENCRYPT flag to determine the direction of the operation, cryptop now includes an 'op' field defining the operation to perform. For digests I've added a new VERIFY digest mode which assumes a digest is present in the input and fails the request with EBADMSG if it doesn't match the internally-computed digest. GCM and CCM already assumed this, and the new AEAD mode requires this for decryption. The new ETA mode now also requires this for decryption, so IPsec and GELI no longer do their own authentication verification. Simple DIGEST operations can also do this, though there are no in-tree consumers. To eventually support some refcounting to close races, the session cookie is now passed to crypto_getop() and clients should no longer set crp_sesssion directly. - Assymteric crypto operation structures should be allocated via crypto_getkreq() and freed via crypto_freekreq(). This permits the crypto layer to track open asym requests and close races with a driver trying to unregister while asym requests are in flight. - crypto_copyback, crypto_copydata, crypto_apply, and crypto_contiguous_subsegment now accept the 'crp' object as the first parameter instead of individual members. This makes it easier to deal with different buffer types in the future as well as separate input and output buffers. It's also simpler for driver writers to use. - bus_dmamap_load_crp() loads a DMA mapping for a crypto buffer. This understands the various types of buffers so that drivers that use DMA do not have to be aware of different buffer types. - Helper routines now exist to build an auth context for HMAC IPAD and OPAD. This reduces some duplicated work among drivers. - Key buffers are now treated as const throughout the framework and in device drivers. However, session key buffers provided when a session is created are expected to remain alive for the duration of the session. - GCM and CCM sessions now only specify a cipher algorithm and a cipher key. The redundant auth information is not needed or used. - For cryptosoft, split up the code a bit such that the 'process' callback now invokes a function pointer in the session. This function pointer is set based on the mode (in effect) though it simplifies a few edge cases that would otherwise be in the switch in 'process'. It does split up GCM vs CCM which I think is more readable even if there is some duplication. - I changed /dev/crypto to support GMAC requests using CRYPTO_AES_NIST_GMAC as an auth algorithm and updated cryptocheck to work with it. - Combined cipher and auth sessions via /dev/crypto now always use ETA mode. The COP_F_CIPHER_FIRST flag is now a no-op that is ignored. This was actually documented as being true in crypto(4) before, but the code had not implemented this before I added the CIPHER_FIRST flag. - I have not yet updated /dev/crypto to be aware of explicit modes for sessions. I will probably do that at some point in the future as well as teach it about IV/nonce and tag lengths for AEAD so we can support all of the NIST KAT tests for GCM and CCM. - I've split up the exising crypto.9 manpage into several pages of which many are written from scratch. - I have converted all drivers and consumers in the tree and verified that they compile, but I have not tested all of them. I have tested the following drivers: - cryptosoft - aesni (AES only) - blake2 - ccr and the following consumers: - cryptodev - IPsec - ktls_ocf - GELI (lightly) I have not tested the following: - ccp - aesni with sha - hifn - kgssapi_krb5 - ubsec - padlock - safe - armv8_crypto (aarch64) - glxsb (i386) - sec (ppc) - cesa (armv7) - cryptocteon (mips64) - nlmsec (mips64) Discussed with: cem Relnotes: yes Sponsored by: Chelsio Communications Differential Revision: https://reviews.freebsd.org/D23677
2020-03-27 18:25:23 +00:00
csp.csp_cipher_key = key;
} else {
Refactor driver and consumer interfaces for OCF (in-kernel crypto). - The linked list of cryptoini structures used in session initialization is replaced with a new flat structure: struct crypto_session_params. This session includes a new mode to define how the other fields should be interpreted. Available modes include: - COMPRESS (for compression/decompression) - CIPHER (for simply encryption/decryption) - DIGEST (computing and verifying digests) - AEAD (combined auth and encryption such as AES-GCM and AES-CCM) - ETA (combined auth and encryption using encrypt-then-authenticate) Additional modes could be added in the future (e.g. if we wanted to support TLS MtE for AES-CBC in the kernel we could add a new mode for that. TLS modes might also affect how AAD is interpreted, etc.) The flat structure also includes the key lengths and algorithms as before. However, code doesn't have to walk the linked list and switch on the algorithm to determine which key is the auth key vs encryption key. The 'csp_auth_*' fields are always used for auth keys and settings and 'csp_cipher_*' for cipher. (Compression algorithms are stored in csp_cipher_alg.) - Drivers no longer register a list of supported algorithms. This doesn't quite work when you factor in modes (e.g. a driver might support both AES-CBC and SHA2-256-HMAC separately but not combined for ETA). Instead, a new 'crypto_probesession' method has been added to the kobj interface for symmteric crypto drivers. This method returns a negative value on success (similar to how device_probe works) and the crypto framework uses this value to pick the "best" driver. There are three constants for hardware (e.g. ccr), accelerated software (e.g. aesni), and plain software (cryptosoft) that give preference in that order. One effect of this is that if you request only hardware when creating a new session, you will no longer get a session using accelerated software. Another effect is that the default setting to disallow software crypto via /dev/crypto now disables accelerated software. Once a driver is chosen, 'crypto_newsession' is invoked as before. - Crypto operations are now solely described by the flat 'cryptop' structure. The linked list of descriptors has been removed. A separate enum has been added to describe the type of data buffer in use instead of using CRYPTO_F_* flags to make it easier to add more types in the future if needed (e.g. wired userspace buffers for zero-copy). It will also make it easier to re-introduce separate input and output buffers (in-kernel TLS would benefit from this). Try to make the flags related to IV handling less insane: - CRYPTO_F_IV_SEPARATE means that the IV is stored in the 'crp_iv' member of the operation structure. If this flag is not set, the IV is stored in the data buffer at the 'crp_iv_start' offset. - CRYPTO_F_IV_GENERATE means that a random IV should be generated and stored into the data buffer. This cannot be used with CRYPTO_F_IV_SEPARATE. If a consumer wants to deal with explicit vs implicit IVs, etc. it can always generate the IV however it needs and store partial IVs in the buffer and the full IV/nonce in crp_iv and set CRYPTO_F_IV_SEPARATE. The layout of the buffer is now described via fields in cryptop. crp_aad_start and crp_aad_length define the boundaries of any AAD. Previously with GCM and CCM you defined an auth crd with this range, but for ETA your auth crd had to span both the AAD and plaintext (and they had to be adjacent). crp_payload_start and crp_payload_length define the boundaries of the plaintext/ciphertext. Modes that only do a single operation (COMPRESS, CIPHER, DIGEST) should only use this region and leave the AAD region empty. If a digest is present (or should be generated), it's starting location is marked by crp_digest_start. Instead of using the CRD_F_ENCRYPT flag to determine the direction of the operation, cryptop now includes an 'op' field defining the operation to perform. For digests I've added a new VERIFY digest mode which assumes a digest is present in the input and fails the request with EBADMSG if it doesn't match the internally-computed digest. GCM and CCM already assumed this, and the new AEAD mode requires this for decryption. The new ETA mode now also requires this for decryption, so IPsec and GELI no longer do their own authentication verification. Simple DIGEST operations can also do this, though there are no in-tree consumers. To eventually support some refcounting to close races, the session cookie is now passed to crypto_getop() and clients should no longer set crp_sesssion directly. - Assymteric crypto operation structures should be allocated via crypto_getkreq() and freed via crypto_freekreq(). This permits the crypto layer to track open asym requests and close races with a driver trying to unregister while asym requests are in flight. - crypto_copyback, crypto_copydata, crypto_apply, and crypto_contiguous_subsegment now accept the 'crp' object as the first parameter instead of individual members. This makes it easier to deal with different buffer types in the future as well as separate input and output buffers. It's also simpler for driver writers to use. - bus_dmamap_load_crp() loads a DMA mapping for a crypto buffer. This understands the various types of buffers so that drivers that use DMA do not have to be aware of different buffer types. - Helper routines now exist to build an auth context for HMAC IPAD and OPAD. This reduces some duplicated work among drivers. - Key buffers are now treated as const throughout the framework and in device drivers. However, session key buffers provided when a session is created are expected to remain alive for the duration of the session. - GCM and CCM sessions now only specify a cipher algorithm and a cipher key. The redundant auth information is not needed or used. - For cryptosoft, split up the code a bit such that the 'process' callback now invokes a function pointer in the session. This function pointer is set based on the mode (in effect) though it simplifies a few edge cases that would otherwise be in the switch in 'process'. It does split up GCM vs CCM which I think is more readable even if there is some duplication. - I changed /dev/crypto to support GMAC requests using CRYPTO_AES_NIST_GMAC as an auth algorithm and updated cryptocheck to work with it. - Combined cipher and auth sessions via /dev/crypto now always use ETA mode. The COP_F_CIPHER_FIRST flag is now a no-op that is ignored. This was actually documented as being true in crypto(4) before, but the code had not implemented this before I added the CIPHER_FIRST flag. - I have not yet updated /dev/crypto to be aware of explicit modes for sessions. I will probably do that at some point in the future as well as teach it about IV/nonce and tag lengths for AEAD so we can support all of the NIST KAT tests for GCM and CCM. - I've split up the exising crypto.9 manpage into several pages of which many are written from scratch. - I have converted all drivers and consumers in the tree and verified that they compile, but I have not tested all of them. I have tested the following drivers: - cryptosoft - aesni (AES only) - blake2 - ccr and the following consumers: - cryptodev - IPsec - ktls_ocf - GELI (lightly) I have not tested the following: - ccp - aesni with sha - hifn - kgssapi_krb5 - ubsec - padlock - safe - armv8_crypto (aarch64) - glxsb (i386) - sec (ppc) - cesa (armv7) - cryptocteon (mips64) - nlmsec (mips64) Discussed with: cem Relnotes: yes Sponsored by: Chelsio Communications Differential Revision: https://reviews.freebsd.org/D23677
2020-03-27 18:25:23 +00:00
key = NULL;
csp.csp_cipher_key = sc->sc_ekey;
}
if (sc->sc_flags & G_ELI_FLAG_AUTH) {
Refactor driver and consumer interfaces for OCF (in-kernel crypto). - The linked list of cryptoini structures used in session initialization is replaced with a new flat structure: struct crypto_session_params. This session includes a new mode to define how the other fields should be interpreted. Available modes include: - COMPRESS (for compression/decompression) - CIPHER (for simply encryption/decryption) - DIGEST (computing and verifying digests) - AEAD (combined auth and encryption such as AES-GCM and AES-CCM) - ETA (combined auth and encryption using encrypt-then-authenticate) Additional modes could be added in the future (e.g. if we wanted to support TLS MtE for AES-CBC in the kernel we could add a new mode for that. TLS modes might also affect how AAD is interpreted, etc.) The flat structure also includes the key lengths and algorithms as before. However, code doesn't have to walk the linked list and switch on the algorithm to determine which key is the auth key vs encryption key. The 'csp_auth_*' fields are always used for auth keys and settings and 'csp_cipher_*' for cipher. (Compression algorithms are stored in csp_cipher_alg.) - Drivers no longer register a list of supported algorithms. This doesn't quite work when you factor in modes (e.g. a driver might support both AES-CBC and SHA2-256-HMAC separately but not combined for ETA). Instead, a new 'crypto_probesession' method has been added to the kobj interface for symmteric crypto drivers. This method returns a negative value on success (similar to how device_probe works) and the crypto framework uses this value to pick the "best" driver. There are three constants for hardware (e.g. ccr), accelerated software (e.g. aesni), and plain software (cryptosoft) that give preference in that order. One effect of this is that if you request only hardware when creating a new session, you will no longer get a session using accelerated software. Another effect is that the default setting to disallow software crypto via /dev/crypto now disables accelerated software. Once a driver is chosen, 'crypto_newsession' is invoked as before. - Crypto operations are now solely described by the flat 'cryptop' structure. The linked list of descriptors has been removed. A separate enum has been added to describe the type of data buffer in use instead of using CRYPTO_F_* flags to make it easier to add more types in the future if needed (e.g. wired userspace buffers for zero-copy). It will also make it easier to re-introduce separate input and output buffers (in-kernel TLS would benefit from this). Try to make the flags related to IV handling less insane: - CRYPTO_F_IV_SEPARATE means that the IV is stored in the 'crp_iv' member of the operation structure. If this flag is not set, the IV is stored in the data buffer at the 'crp_iv_start' offset. - CRYPTO_F_IV_GENERATE means that a random IV should be generated and stored into the data buffer. This cannot be used with CRYPTO_F_IV_SEPARATE. If a consumer wants to deal with explicit vs implicit IVs, etc. it can always generate the IV however it needs and store partial IVs in the buffer and the full IV/nonce in crp_iv and set CRYPTO_F_IV_SEPARATE. The layout of the buffer is now described via fields in cryptop. crp_aad_start and crp_aad_length define the boundaries of any AAD. Previously with GCM and CCM you defined an auth crd with this range, but for ETA your auth crd had to span both the AAD and plaintext (and they had to be adjacent). crp_payload_start and crp_payload_length define the boundaries of the plaintext/ciphertext. Modes that only do a single operation (COMPRESS, CIPHER, DIGEST) should only use this region and leave the AAD region empty. If a digest is present (or should be generated), it's starting location is marked by crp_digest_start. Instead of using the CRD_F_ENCRYPT flag to determine the direction of the operation, cryptop now includes an 'op' field defining the operation to perform. For digests I've added a new VERIFY digest mode which assumes a digest is present in the input and fails the request with EBADMSG if it doesn't match the internally-computed digest. GCM and CCM already assumed this, and the new AEAD mode requires this for decryption. The new ETA mode now also requires this for decryption, so IPsec and GELI no longer do their own authentication verification. Simple DIGEST operations can also do this, though there are no in-tree consumers. To eventually support some refcounting to close races, the session cookie is now passed to crypto_getop() and clients should no longer set crp_sesssion directly. - Assymteric crypto operation structures should be allocated via crypto_getkreq() and freed via crypto_freekreq(). This permits the crypto layer to track open asym requests and close races with a driver trying to unregister while asym requests are in flight. - crypto_copyback, crypto_copydata, crypto_apply, and crypto_contiguous_subsegment now accept the 'crp' object as the first parameter instead of individual members. This makes it easier to deal with different buffer types in the future as well as separate input and output buffers. It's also simpler for driver writers to use. - bus_dmamap_load_crp() loads a DMA mapping for a crypto buffer. This understands the various types of buffers so that drivers that use DMA do not have to be aware of different buffer types. - Helper routines now exist to build an auth context for HMAC IPAD and OPAD. This reduces some duplicated work among drivers. - Key buffers are now treated as const throughout the framework and in device drivers. However, session key buffers provided when a session is created are expected to remain alive for the duration of the session. - GCM and CCM sessions now only specify a cipher algorithm and a cipher key. The redundant auth information is not needed or used. - For cryptosoft, split up the code a bit such that the 'process' callback now invokes a function pointer in the session. This function pointer is set based on the mode (in effect) though it simplifies a few edge cases that would otherwise be in the switch in 'process'. It does split up GCM vs CCM which I think is more readable even if there is some duplication. - I changed /dev/crypto to support GMAC requests using CRYPTO_AES_NIST_GMAC as an auth algorithm and updated cryptocheck to work with it. - Combined cipher and auth sessions via /dev/crypto now always use ETA mode. The COP_F_CIPHER_FIRST flag is now a no-op that is ignored. This was actually documented as being true in crypto(4) before, but the code had not implemented this before I added the CIPHER_FIRST flag. - I have not yet updated /dev/crypto to be aware of explicit modes for sessions. I will probably do that at some point in the future as well as teach it about IV/nonce and tag lengths for AEAD so we can support all of the NIST KAT tests for GCM and CCM. - I've split up the exising crypto.9 manpage into several pages of which many are written from scratch. - I have converted all drivers and consumers in the tree and verified that they compile, but I have not tested all of them. I have tested the following drivers: - cryptosoft - aesni (AES only) - blake2 - ccr and the following consumers: - cryptodev - IPsec - ktls_ocf - GELI (lightly) I have not tested the following: - ccp - aesni with sha - hifn - kgssapi_krb5 - ubsec - padlock - safe - armv8_crypto (aarch64) - glxsb (i386) - sec (ppc) - cesa (armv7) - cryptocteon (mips64) - nlmsec (mips64) Discussed with: cem Relnotes: yes Sponsored by: Chelsio Communications Differential Revision: https://reviews.freebsd.org/D23677
2020-03-27 18:25:23 +00:00
csp.csp_mode = CSP_MODE_ETA;
csp.csp_auth_alg = sc->sc_aalgo;
csp.csp_auth_klen = G_ELI_AUTH_SECKEYLEN;
}
switch (sc->sc_crypto) {
case G_ELI_CRYPTO_SW_ACCEL:
case G_ELI_CRYPTO_SW:
Refactor driver and consumer interfaces for OCF (in-kernel crypto). - The linked list of cryptoini structures used in session initialization is replaced with a new flat structure: struct crypto_session_params. This session includes a new mode to define how the other fields should be interpreted. Available modes include: - COMPRESS (for compression/decompression) - CIPHER (for simply encryption/decryption) - DIGEST (computing and verifying digests) - AEAD (combined auth and encryption such as AES-GCM and AES-CCM) - ETA (combined auth and encryption using encrypt-then-authenticate) Additional modes could be added in the future (e.g. if we wanted to support TLS MtE for AES-CBC in the kernel we could add a new mode for that. TLS modes might also affect how AAD is interpreted, etc.) The flat structure also includes the key lengths and algorithms as before. However, code doesn't have to walk the linked list and switch on the algorithm to determine which key is the auth key vs encryption key. The 'csp_auth_*' fields are always used for auth keys and settings and 'csp_cipher_*' for cipher. (Compression algorithms are stored in csp_cipher_alg.) - Drivers no longer register a list of supported algorithms. This doesn't quite work when you factor in modes (e.g. a driver might support both AES-CBC and SHA2-256-HMAC separately but not combined for ETA). Instead, a new 'crypto_probesession' method has been added to the kobj interface for symmteric crypto drivers. This method returns a negative value on success (similar to how device_probe works) and the crypto framework uses this value to pick the "best" driver. There are three constants for hardware (e.g. ccr), accelerated software (e.g. aesni), and plain software (cryptosoft) that give preference in that order. One effect of this is that if you request only hardware when creating a new session, you will no longer get a session using accelerated software. Another effect is that the default setting to disallow software crypto via /dev/crypto now disables accelerated software. Once a driver is chosen, 'crypto_newsession' is invoked as before. - Crypto operations are now solely described by the flat 'cryptop' structure. The linked list of descriptors has been removed. A separate enum has been added to describe the type of data buffer in use instead of using CRYPTO_F_* flags to make it easier to add more types in the future if needed (e.g. wired userspace buffers for zero-copy). It will also make it easier to re-introduce separate input and output buffers (in-kernel TLS would benefit from this). Try to make the flags related to IV handling less insane: - CRYPTO_F_IV_SEPARATE means that the IV is stored in the 'crp_iv' member of the operation structure. If this flag is not set, the IV is stored in the data buffer at the 'crp_iv_start' offset. - CRYPTO_F_IV_GENERATE means that a random IV should be generated and stored into the data buffer. This cannot be used with CRYPTO_F_IV_SEPARATE. If a consumer wants to deal with explicit vs implicit IVs, etc. it can always generate the IV however it needs and store partial IVs in the buffer and the full IV/nonce in crp_iv and set CRYPTO_F_IV_SEPARATE. The layout of the buffer is now described via fields in cryptop. crp_aad_start and crp_aad_length define the boundaries of any AAD. Previously with GCM and CCM you defined an auth crd with this range, but for ETA your auth crd had to span both the AAD and plaintext (and they had to be adjacent). crp_payload_start and crp_payload_length define the boundaries of the plaintext/ciphertext. Modes that only do a single operation (COMPRESS, CIPHER, DIGEST) should only use this region and leave the AAD region empty. If a digest is present (or should be generated), it's starting location is marked by crp_digest_start. Instead of using the CRD_F_ENCRYPT flag to determine the direction of the operation, cryptop now includes an 'op' field defining the operation to perform. For digests I've added a new VERIFY digest mode which assumes a digest is present in the input and fails the request with EBADMSG if it doesn't match the internally-computed digest. GCM and CCM already assumed this, and the new AEAD mode requires this for decryption. The new ETA mode now also requires this for decryption, so IPsec and GELI no longer do their own authentication verification. Simple DIGEST operations can also do this, though there are no in-tree consumers. To eventually support some refcounting to close races, the session cookie is now passed to crypto_getop() and clients should no longer set crp_sesssion directly. - Assymteric crypto operation structures should be allocated via crypto_getkreq() and freed via crypto_freekreq(). This permits the crypto layer to track open asym requests and close races with a driver trying to unregister while asym requests are in flight. - crypto_copyback, crypto_copydata, crypto_apply, and crypto_contiguous_subsegment now accept the 'crp' object as the first parameter instead of individual members. This makes it easier to deal with different buffer types in the future as well as separate input and output buffers. It's also simpler for driver writers to use. - bus_dmamap_load_crp() loads a DMA mapping for a crypto buffer. This understands the various types of buffers so that drivers that use DMA do not have to be aware of different buffer types. - Helper routines now exist to build an auth context for HMAC IPAD and OPAD. This reduces some duplicated work among drivers. - Key buffers are now treated as const throughout the framework and in device drivers. However, session key buffers provided when a session is created are expected to remain alive for the duration of the session. - GCM and CCM sessions now only specify a cipher algorithm and a cipher key. The redundant auth information is not needed or used. - For cryptosoft, split up the code a bit such that the 'process' callback now invokes a function pointer in the session. This function pointer is set based on the mode (in effect) though it simplifies a few edge cases that would otherwise be in the switch in 'process'. It does split up GCM vs CCM which I think is more readable even if there is some duplication. - I changed /dev/crypto to support GMAC requests using CRYPTO_AES_NIST_GMAC as an auth algorithm and updated cryptocheck to work with it. - Combined cipher and auth sessions via /dev/crypto now always use ETA mode. The COP_F_CIPHER_FIRST flag is now a no-op that is ignored. This was actually documented as being true in crypto(4) before, but the code had not implemented this before I added the CIPHER_FIRST flag. - I have not yet updated /dev/crypto to be aware of explicit modes for sessions. I will probably do that at some point in the future as well as teach it about IV/nonce and tag lengths for AEAD so we can support all of the NIST KAT tests for GCM and CCM. - I've split up the exising crypto.9 manpage into several pages of which many are written from scratch. - I have converted all drivers and consumers in the tree and verified that they compile, but I have not tested all of them. I have tested the following drivers: - cryptosoft - aesni (AES only) - blake2 - ccr and the following consumers: - cryptodev - IPsec - ktls_ocf - GELI (lightly) I have not tested the following: - ccp - aesni with sha - hifn - kgssapi_krb5 - ubsec - padlock - safe - armv8_crypto (aarch64) - glxsb (i386) - sec (ppc) - cesa (armv7) - cryptocteon (mips64) - nlmsec (mips64) Discussed with: cem Relnotes: yes Sponsored by: Chelsio Communications Differential Revision: https://reviews.freebsd.org/D23677
2020-03-27 18:25:23 +00:00
error = crypto_newsession(&wr->w_sid, &csp,
CRYPTOCAP_F_SOFTWARE);
break;
case G_ELI_CRYPTO_HW:
Refactor driver and consumer interfaces for OCF (in-kernel crypto). - The linked list of cryptoini structures used in session initialization is replaced with a new flat structure: struct crypto_session_params. This session includes a new mode to define how the other fields should be interpreted. Available modes include: - COMPRESS (for compression/decompression) - CIPHER (for simply encryption/decryption) - DIGEST (computing and verifying digests) - AEAD (combined auth and encryption such as AES-GCM and AES-CCM) - ETA (combined auth and encryption using encrypt-then-authenticate) Additional modes could be added in the future (e.g. if we wanted to support TLS MtE for AES-CBC in the kernel we could add a new mode for that. TLS modes might also affect how AAD is interpreted, etc.) The flat structure also includes the key lengths and algorithms as before. However, code doesn't have to walk the linked list and switch on the algorithm to determine which key is the auth key vs encryption key. The 'csp_auth_*' fields are always used for auth keys and settings and 'csp_cipher_*' for cipher. (Compression algorithms are stored in csp_cipher_alg.) - Drivers no longer register a list of supported algorithms. This doesn't quite work when you factor in modes (e.g. a driver might support both AES-CBC and SHA2-256-HMAC separately but not combined for ETA). Instead, a new 'crypto_probesession' method has been added to the kobj interface for symmteric crypto drivers. This method returns a negative value on success (similar to how device_probe works) and the crypto framework uses this value to pick the "best" driver. There are three constants for hardware (e.g. ccr), accelerated software (e.g. aesni), and plain software (cryptosoft) that give preference in that order. One effect of this is that if you request only hardware when creating a new session, you will no longer get a session using accelerated software. Another effect is that the default setting to disallow software crypto via /dev/crypto now disables accelerated software. Once a driver is chosen, 'crypto_newsession' is invoked as before. - Crypto operations are now solely described by the flat 'cryptop' structure. The linked list of descriptors has been removed. A separate enum has been added to describe the type of data buffer in use instead of using CRYPTO_F_* flags to make it easier to add more types in the future if needed (e.g. wired userspace buffers for zero-copy). It will also make it easier to re-introduce separate input and output buffers (in-kernel TLS would benefit from this). Try to make the flags related to IV handling less insane: - CRYPTO_F_IV_SEPARATE means that the IV is stored in the 'crp_iv' member of the operation structure. If this flag is not set, the IV is stored in the data buffer at the 'crp_iv_start' offset. - CRYPTO_F_IV_GENERATE means that a random IV should be generated and stored into the data buffer. This cannot be used with CRYPTO_F_IV_SEPARATE. If a consumer wants to deal with explicit vs implicit IVs, etc. it can always generate the IV however it needs and store partial IVs in the buffer and the full IV/nonce in crp_iv and set CRYPTO_F_IV_SEPARATE. The layout of the buffer is now described via fields in cryptop. crp_aad_start and crp_aad_length define the boundaries of any AAD. Previously with GCM and CCM you defined an auth crd with this range, but for ETA your auth crd had to span both the AAD and plaintext (and they had to be adjacent). crp_payload_start and crp_payload_length define the boundaries of the plaintext/ciphertext. Modes that only do a single operation (COMPRESS, CIPHER, DIGEST) should only use this region and leave the AAD region empty. If a digest is present (or should be generated), it's starting location is marked by crp_digest_start. Instead of using the CRD_F_ENCRYPT flag to determine the direction of the operation, cryptop now includes an 'op' field defining the operation to perform. For digests I've added a new VERIFY digest mode which assumes a digest is present in the input and fails the request with EBADMSG if it doesn't match the internally-computed digest. GCM and CCM already assumed this, and the new AEAD mode requires this for decryption. The new ETA mode now also requires this for decryption, so IPsec and GELI no longer do their own authentication verification. Simple DIGEST operations can also do this, though there are no in-tree consumers. To eventually support some refcounting to close races, the session cookie is now passed to crypto_getop() and clients should no longer set crp_sesssion directly. - Assymteric crypto operation structures should be allocated via crypto_getkreq() and freed via crypto_freekreq(). This permits the crypto layer to track open asym requests and close races with a driver trying to unregister while asym requests are in flight. - crypto_copyback, crypto_copydata, crypto_apply, and crypto_contiguous_subsegment now accept the 'crp' object as the first parameter instead of individual members. This makes it easier to deal with different buffer types in the future as well as separate input and output buffers. It's also simpler for driver writers to use. - bus_dmamap_load_crp() loads a DMA mapping for a crypto buffer. This understands the various types of buffers so that drivers that use DMA do not have to be aware of different buffer types. - Helper routines now exist to build an auth context for HMAC IPAD and OPAD. This reduces some duplicated work among drivers. - Key buffers are now treated as const throughout the framework and in device drivers. However, session key buffers provided when a session is created are expected to remain alive for the duration of the session. - GCM and CCM sessions now only specify a cipher algorithm and a cipher key. The redundant auth information is not needed or used. - For cryptosoft, split up the code a bit such that the 'process' callback now invokes a function pointer in the session. This function pointer is set based on the mode (in effect) though it simplifies a few edge cases that would otherwise be in the switch in 'process'. It does split up GCM vs CCM which I think is more readable even if there is some duplication. - I changed /dev/crypto to support GMAC requests using CRYPTO_AES_NIST_GMAC as an auth algorithm and updated cryptocheck to work with it. - Combined cipher and auth sessions via /dev/crypto now always use ETA mode. The COP_F_CIPHER_FIRST flag is now a no-op that is ignored. This was actually documented as being true in crypto(4) before, but the code had not implemented this before I added the CIPHER_FIRST flag. - I have not yet updated /dev/crypto to be aware of explicit modes for sessions. I will probably do that at some point in the future as well as teach it about IV/nonce and tag lengths for AEAD so we can support all of the NIST KAT tests for GCM and CCM. - I've split up the exising crypto.9 manpage into several pages of which many are written from scratch. - I have converted all drivers and consumers in the tree and verified that they compile, but I have not tested all of them. I have tested the following drivers: - cryptosoft - aesni (AES only) - blake2 - ccr and the following consumers: - cryptodev - IPsec - ktls_ocf - GELI (lightly) I have not tested the following: - ccp - aesni with sha - hifn - kgssapi_krb5 - ubsec - padlock - safe - armv8_crypto (aarch64) - glxsb (i386) - sec (ppc) - cesa (armv7) - cryptocteon (mips64) - nlmsec (mips64) Discussed with: cem Relnotes: yes Sponsored by: Chelsio Communications Differential Revision: https://reviews.freebsd.org/D23677
2020-03-27 18:25:23 +00:00
error = crypto_newsession(&wr->w_sid, &csp,
CRYPTOCAP_F_HARDWARE);
break;
case G_ELI_CRYPTO_UNKNOWN:
Refactor driver and consumer interfaces for OCF (in-kernel crypto). - The linked list of cryptoini structures used in session initialization is replaced with a new flat structure: struct crypto_session_params. This session includes a new mode to define how the other fields should be interpreted. Available modes include: - COMPRESS (for compression/decompression) - CIPHER (for simply encryption/decryption) - DIGEST (computing and verifying digests) - AEAD (combined auth and encryption such as AES-GCM and AES-CCM) - ETA (combined auth and encryption using encrypt-then-authenticate) Additional modes could be added in the future (e.g. if we wanted to support TLS MtE for AES-CBC in the kernel we could add a new mode for that. TLS modes might also affect how AAD is interpreted, etc.) The flat structure also includes the key lengths and algorithms as before. However, code doesn't have to walk the linked list and switch on the algorithm to determine which key is the auth key vs encryption key. The 'csp_auth_*' fields are always used for auth keys and settings and 'csp_cipher_*' for cipher. (Compression algorithms are stored in csp_cipher_alg.) - Drivers no longer register a list of supported algorithms. This doesn't quite work when you factor in modes (e.g. a driver might support both AES-CBC and SHA2-256-HMAC separately but not combined for ETA). Instead, a new 'crypto_probesession' method has been added to the kobj interface for symmteric crypto drivers. This method returns a negative value on success (similar to how device_probe works) and the crypto framework uses this value to pick the "best" driver. There are three constants for hardware (e.g. ccr), accelerated software (e.g. aesni), and plain software (cryptosoft) that give preference in that order. One effect of this is that if you request only hardware when creating a new session, you will no longer get a session using accelerated software. Another effect is that the default setting to disallow software crypto via /dev/crypto now disables accelerated software. Once a driver is chosen, 'crypto_newsession' is invoked as before. - Crypto operations are now solely described by the flat 'cryptop' structure. The linked list of descriptors has been removed. A separate enum has been added to describe the type of data buffer in use instead of using CRYPTO_F_* flags to make it easier to add more types in the future if needed (e.g. wired userspace buffers for zero-copy). It will also make it easier to re-introduce separate input and output buffers (in-kernel TLS would benefit from this). Try to make the flags related to IV handling less insane: - CRYPTO_F_IV_SEPARATE means that the IV is stored in the 'crp_iv' member of the operation structure. If this flag is not set, the IV is stored in the data buffer at the 'crp_iv_start' offset. - CRYPTO_F_IV_GENERATE means that a random IV should be generated and stored into the data buffer. This cannot be used with CRYPTO_F_IV_SEPARATE. If a consumer wants to deal with explicit vs implicit IVs, etc. it can always generate the IV however it needs and store partial IVs in the buffer and the full IV/nonce in crp_iv and set CRYPTO_F_IV_SEPARATE. The layout of the buffer is now described via fields in cryptop. crp_aad_start and crp_aad_length define the boundaries of any AAD. Previously with GCM and CCM you defined an auth crd with this range, but for ETA your auth crd had to span both the AAD and plaintext (and they had to be adjacent). crp_payload_start and crp_payload_length define the boundaries of the plaintext/ciphertext. Modes that only do a single operation (COMPRESS, CIPHER, DIGEST) should only use this region and leave the AAD region empty. If a digest is present (or should be generated), it's starting location is marked by crp_digest_start. Instead of using the CRD_F_ENCRYPT flag to determine the direction of the operation, cryptop now includes an 'op' field defining the operation to perform. For digests I've added a new VERIFY digest mode which assumes a digest is present in the input and fails the request with EBADMSG if it doesn't match the internally-computed digest. GCM and CCM already assumed this, and the new AEAD mode requires this for decryption. The new ETA mode now also requires this for decryption, so IPsec and GELI no longer do their own authentication verification. Simple DIGEST operations can also do this, though there are no in-tree consumers. To eventually support some refcounting to close races, the session cookie is now passed to crypto_getop() and clients should no longer set crp_sesssion directly. - Assymteric crypto operation structures should be allocated via crypto_getkreq() and freed via crypto_freekreq(). This permits the crypto layer to track open asym requests and close races with a driver trying to unregister while asym requests are in flight. - crypto_copyback, crypto_copydata, crypto_apply, and crypto_contiguous_subsegment now accept the 'crp' object as the first parameter instead of individual members. This makes it easier to deal with different buffer types in the future as well as separate input and output buffers. It's also simpler for driver writers to use. - bus_dmamap_load_crp() loads a DMA mapping for a crypto buffer. This understands the various types of buffers so that drivers that use DMA do not have to be aware of different buffer types. - Helper routines now exist to build an auth context for HMAC IPAD and OPAD. This reduces some duplicated work among drivers. - Key buffers are now treated as const throughout the framework and in device drivers. However, session key buffers provided when a session is created are expected to remain alive for the duration of the session. - GCM and CCM sessions now only specify a cipher algorithm and a cipher key. The redundant auth information is not needed or used. - For cryptosoft, split up the code a bit such that the 'process' callback now invokes a function pointer in the session. This function pointer is set based on the mode (in effect) though it simplifies a few edge cases that would otherwise be in the switch in 'process'. It does split up GCM vs CCM which I think is more readable even if there is some duplication. - I changed /dev/crypto to support GMAC requests using CRYPTO_AES_NIST_GMAC as an auth algorithm and updated cryptocheck to work with it. - Combined cipher and auth sessions via /dev/crypto now always use ETA mode. The COP_F_CIPHER_FIRST flag is now a no-op that is ignored. This was actually documented as being true in crypto(4) before, but the code had not implemented this before I added the CIPHER_FIRST flag. - I have not yet updated /dev/crypto to be aware of explicit modes for sessions. I will probably do that at some point in the future as well as teach it about IV/nonce and tag lengths for AEAD so we can support all of the NIST KAT tests for GCM and CCM. - I've split up the exising crypto.9 manpage into several pages of which many are written from scratch. - I have converted all drivers and consumers in the tree and verified that they compile, but I have not tested all of them. I have tested the following drivers: - cryptosoft - aesni (AES only) - blake2 - ccr and the following consumers: - cryptodev - IPsec - ktls_ocf - GELI (lightly) I have not tested the following: - ccp - aesni with sha - hifn - kgssapi_krb5 - ubsec - padlock - safe - armv8_crypto (aarch64) - glxsb (i386) - sec (ppc) - cesa (armv7) - cryptocteon (mips64) - nlmsec (mips64) Discussed with: cem Relnotes: yes Sponsored by: Chelsio Communications Differential Revision: https://reviews.freebsd.org/D23677
2020-03-27 18:25:23 +00:00
error = crypto_newsession(&wr->w_sid, &csp,
CRYPTOCAP_F_HARDWARE | CRYPTOCAP_F_SOFTWARE);
if (error == 0) {
caps = crypto_ses2caps(wr->w_sid);
if (caps & CRYPTOCAP_F_HARDWARE)
new_crypto = G_ELI_CRYPTO_HW;
else if (caps & CRYPTOCAP_F_ACCEL_SOFTWARE)
new_crypto = G_ELI_CRYPTO_SW_ACCEL;
else
new_crypto = G_ELI_CRYPTO_SW;
mtx_lock(&sc->sc_queue_mtx);
if (sc->sc_crypto == G_ELI_CRYPTO_UNKNOWN)
sc->sc_crypto = new_crypto;
mtx_unlock(&sc->sc_queue_mtx);
}
break;
default:
panic("%s: invalid condition", __func__);
}
Refactor driver and consumer interfaces for OCF (in-kernel crypto). - The linked list of cryptoini structures used in session initialization is replaced with a new flat structure: struct crypto_session_params. This session includes a new mode to define how the other fields should be interpreted. Available modes include: - COMPRESS (for compression/decompression) - CIPHER (for simply encryption/decryption) - DIGEST (computing and verifying digests) - AEAD (combined auth and encryption such as AES-GCM and AES-CCM) - ETA (combined auth and encryption using encrypt-then-authenticate) Additional modes could be added in the future (e.g. if we wanted to support TLS MtE for AES-CBC in the kernel we could add a new mode for that. TLS modes might also affect how AAD is interpreted, etc.) The flat structure also includes the key lengths and algorithms as before. However, code doesn't have to walk the linked list and switch on the algorithm to determine which key is the auth key vs encryption key. The 'csp_auth_*' fields are always used for auth keys and settings and 'csp_cipher_*' for cipher. (Compression algorithms are stored in csp_cipher_alg.) - Drivers no longer register a list of supported algorithms. This doesn't quite work when you factor in modes (e.g. a driver might support both AES-CBC and SHA2-256-HMAC separately but not combined for ETA). Instead, a new 'crypto_probesession' method has been added to the kobj interface for symmteric crypto drivers. This method returns a negative value on success (similar to how device_probe works) and the crypto framework uses this value to pick the "best" driver. There are three constants for hardware (e.g. ccr), accelerated software (e.g. aesni), and plain software (cryptosoft) that give preference in that order. One effect of this is that if you request only hardware when creating a new session, you will no longer get a session using accelerated software. Another effect is that the default setting to disallow software crypto via /dev/crypto now disables accelerated software. Once a driver is chosen, 'crypto_newsession' is invoked as before. - Crypto operations are now solely described by the flat 'cryptop' structure. The linked list of descriptors has been removed. A separate enum has been added to describe the type of data buffer in use instead of using CRYPTO_F_* flags to make it easier to add more types in the future if needed (e.g. wired userspace buffers for zero-copy). It will also make it easier to re-introduce separate input and output buffers (in-kernel TLS would benefit from this). Try to make the flags related to IV handling less insane: - CRYPTO_F_IV_SEPARATE means that the IV is stored in the 'crp_iv' member of the operation structure. If this flag is not set, the IV is stored in the data buffer at the 'crp_iv_start' offset. - CRYPTO_F_IV_GENERATE means that a random IV should be generated and stored into the data buffer. This cannot be used with CRYPTO_F_IV_SEPARATE. If a consumer wants to deal with explicit vs implicit IVs, etc. it can always generate the IV however it needs and store partial IVs in the buffer and the full IV/nonce in crp_iv and set CRYPTO_F_IV_SEPARATE. The layout of the buffer is now described via fields in cryptop. crp_aad_start and crp_aad_length define the boundaries of any AAD. Previously with GCM and CCM you defined an auth crd with this range, but for ETA your auth crd had to span both the AAD and plaintext (and they had to be adjacent). crp_payload_start and crp_payload_length define the boundaries of the plaintext/ciphertext. Modes that only do a single operation (COMPRESS, CIPHER, DIGEST) should only use this region and leave the AAD region empty. If a digest is present (or should be generated), it's starting location is marked by crp_digest_start. Instead of using the CRD_F_ENCRYPT flag to determine the direction of the operation, cryptop now includes an 'op' field defining the operation to perform. For digests I've added a new VERIFY digest mode which assumes a digest is present in the input and fails the request with EBADMSG if it doesn't match the internally-computed digest. GCM and CCM already assumed this, and the new AEAD mode requires this for decryption. The new ETA mode now also requires this for decryption, so IPsec and GELI no longer do their own authentication verification. Simple DIGEST operations can also do this, though there are no in-tree consumers. To eventually support some refcounting to close races, the session cookie is now passed to crypto_getop() and clients should no longer set crp_sesssion directly. - Assymteric crypto operation structures should be allocated via crypto_getkreq() and freed via crypto_freekreq(). This permits the crypto layer to track open asym requests and close races with a driver trying to unregister while asym requests are in flight. - crypto_copyback, crypto_copydata, crypto_apply, and crypto_contiguous_subsegment now accept the 'crp' object as the first parameter instead of individual members. This makes it easier to deal with different buffer types in the future as well as separate input and output buffers. It's also simpler for driver writers to use. - bus_dmamap_load_crp() loads a DMA mapping for a crypto buffer. This understands the various types of buffers so that drivers that use DMA do not have to be aware of different buffer types. - Helper routines now exist to build an auth context for HMAC IPAD and OPAD. This reduces some duplicated work among drivers. - Key buffers are now treated as const throughout the framework and in device drivers. However, session key buffers provided when a session is created are expected to remain alive for the duration of the session. - GCM and CCM sessions now only specify a cipher algorithm and a cipher key. The redundant auth information is not needed or used. - For cryptosoft, split up the code a bit such that the 'process' callback now invokes a function pointer in the session. This function pointer is set based on the mode (in effect) though it simplifies a few edge cases that would otherwise be in the switch in 'process'. It does split up GCM vs CCM which I think is more readable even if there is some duplication. - I changed /dev/crypto to support GMAC requests using CRYPTO_AES_NIST_GMAC as an auth algorithm and updated cryptocheck to work with it. - Combined cipher and auth sessions via /dev/crypto now always use ETA mode. The COP_F_CIPHER_FIRST flag is now a no-op that is ignored. This was actually documented as being true in crypto(4) before, but the code had not implemented this before I added the CIPHER_FIRST flag. - I have not yet updated /dev/crypto to be aware of explicit modes for sessions. I will probably do that at some point in the future as well as teach it about IV/nonce and tag lengths for AEAD so we can support all of the NIST KAT tests for GCM and CCM. - I've split up the exising crypto.9 manpage into several pages of which many are written from scratch. - I have converted all drivers and consumers in the tree and verified that they compile, but I have not tested all of them. I have tested the following drivers: - cryptosoft - aesni (AES only) - blake2 - ccr and the following consumers: - cryptodev - IPsec - ktls_ocf - GELI (lightly) I have not tested the following: - ccp - aesni with sha - hifn - kgssapi_krb5 - ubsec - padlock - safe - armv8_crypto (aarch64) - glxsb (i386) - sec (ppc) - cesa (armv7) - cryptocteon (mips64) - nlmsec (mips64) Discussed with: cem Relnotes: yes Sponsored by: Chelsio Communications Differential Revision: https://reviews.freebsd.org/D23677
2020-03-27 18:25:23 +00:00
if ((sc->sc_flags & G_ELI_FLAG_FIRST_KEY) != 0) {
if (error)
g_eli_key_drop(sc, key);
else
wr->w_first_key = key;
}
return (error);
}
static void
g_eli_freesession(struct g_eli_worker *wr)
{
Refactor driver and consumer interfaces for OCF (in-kernel crypto). - The linked list of cryptoini structures used in session initialization is replaced with a new flat structure: struct crypto_session_params. This session includes a new mode to define how the other fields should be interpreted. Available modes include: - COMPRESS (for compression/decompression) - CIPHER (for simply encryption/decryption) - DIGEST (computing and verifying digests) - AEAD (combined auth and encryption such as AES-GCM and AES-CCM) - ETA (combined auth and encryption using encrypt-then-authenticate) Additional modes could be added in the future (e.g. if we wanted to support TLS MtE for AES-CBC in the kernel we could add a new mode for that. TLS modes might also affect how AAD is interpreted, etc.) The flat structure also includes the key lengths and algorithms as before. However, code doesn't have to walk the linked list and switch on the algorithm to determine which key is the auth key vs encryption key. The 'csp_auth_*' fields are always used for auth keys and settings and 'csp_cipher_*' for cipher. (Compression algorithms are stored in csp_cipher_alg.) - Drivers no longer register a list of supported algorithms. This doesn't quite work when you factor in modes (e.g. a driver might support both AES-CBC and SHA2-256-HMAC separately but not combined for ETA). Instead, a new 'crypto_probesession' method has been added to the kobj interface for symmteric crypto drivers. This method returns a negative value on success (similar to how device_probe works) and the crypto framework uses this value to pick the "best" driver. There are three constants for hardware (e.g. ccr), accelerated software (e.g. aesni), and plain software (cryptosoft) that give preference in that order. One effect of this is that if you request only hardware when creating a new session, you will no longer get a session using accelerated software. Another effect is that the default setting to disallow software crypto via /dev/crypto now disables accelerated software. Once a driver is chosen, 'crypto_newsession' is invoked as before. - Crypto operations are now solely described by the flat 'cryptop' structure. The linked list of descriptors has been removed. A separate enum has been added to describe the type of data buffer in use instead of using CRYPTO_F_* flags to make it easier to add more types in the future if needed (e.g. wired userspace buffers for zero-copy). It will also make it easier to re-introduce separate input and output buffers (in-kernel TLS would benefit from this). Try to make the flags related to IV handling less insane: - CRYPTO_F_IV_SEPARATE means that the IV is stored in the 'crp_iv' member of the operation structure. If this flag is not set, the IV is stored in the data buffer at the 'crp_iv_start' offset. - CRYPTO_F_IV_GENERATE means that a random IV should be generated and stored into the data buffer. This cannot be used with CRYPTO_F_IV_SEPARATE. If a consumer wants to deal with explicit vs implicit IVs, etc. it can always generate the IV however it needs and store partial IVs in the buffer and the full IV/nonce in crp_iv and set CRYPTO_F_IV_SEPARATE. The layout of the buffer is now described via fields in cryptop. crp_aad_start and crp_aad_length define the boundaries of any AAD. Previously with GCM and CCM you defined an auth crd with this range, but for ETA your auth crd had to span both the AAD and plaintext (and they had to be adjacent). crp_payload_start and crp_payload_length define the boundaries of the plaintext/ciphertext. Modes that only do a single operation (COMPRESS, CIPHER, DIGEST) should only use this region and leave the AAD region empty. If a digest is present (or should be generated), it's starting location is marked by crp_digest_start. Instead of using the CRD_F_ENCRYPT flag to determine the direction of the operation, cryptop now includes an 'op' field defining the operation to perform. For digests I've added a new VERIFY digest mode which assumes a digest is present in the input and fails the request with EBADMSG if it doesn't match the internally-computed digest. GCM and CCM already assumed this, and the new AEAD mode requires this for decryption. The new ETA mode now also requires this for decryption, so IPsec and GELI no longer do their own authentication verification. Simple DIGEST operations can also do this, though there are no in-tree consumers. To eventually support some refcounting to close races, the session cookie is now passed to crypto_getop() and clients should no longer set crp_sesssion directly. - Assymteric crypto operation structures should be allocated via crypto_getkreq() and freed via crypto_freekreq(). This permits the crypto layer to track open asym requests and close races with a driver trying to unregister while asym requests are in flight. - crypto_copyback, crypto_copydata, crypto_apply, and crypto_contiguous_subsegment now accept the 'crp' object as the first parameter instead of individual members. This makes it easier to deal with different buffer types in the future as well as separate input and output buffers. It's also simpler for driver writers to use. - bus_dmamap_load_crp() loads a DMA mapping for a crypto buffer. This understands the various types of buffers so that drivers that use DMA do not have to be aware of different buffer types. - Helper routines now exist to build an auth context for HMAC IPAD and OPAD. This reduces some duplicated work among drivers. - Key buffers are now treated as const throughout the framework and in device drivers. However, session key buffers provided when a session is created are expected to remain alive for the duration of the session. - GCM and CCM sessions now only specify a cipher algorithm and a cipher key. The redundant auth information is not needed or used. - For cryptosoft, split up the code a bit such that the 'process' callback now invokes a function pointer in the session. This function pointer is set based on the mode (in effect) though it simplifies a few edge cases that would otherwise be in the switch in 'process'. It does split up GCM vs CCM which I think is more readable even if there is some duplication. - I changed /dev/crypto to support GMAC requests using CRYPTO_AES_NIST_GMAC as an auth algorithm and updated cryptocheck to work with it. - Combined cipher and auth sessions via /dev/crypto now always use ETA mode. The COP_F_CIPHER_FIRST flag is now a no-op that is ignored. This was actually documented as being true in crypto(4) before, but the code had not implemented this before I added the CIPHER_FIRST flag. - I have not yet updated /dev/crypto to be aware of explicit modes for sessions. I will probably do that at some point in the future as well as teach it about IV/nonce and tag lengths for AEAD so we can support all of the NIST KAT tests for GCM and CCM. - I've split up the exising crypto.9 manpage into several pages of which many are written from scratch. - I have converted all drivers and consumers in the tree and verified that they compile, but I have not tested all of them. I have tested the following drivers: - cryptosoft - aesni (AES only) - blake2 - ccr and the following consumers: - cryptodev - IPsec - ktls_ocf - GELI (lightly) I have not tested the following: - ccp - aesni with sha - hifn - kgssapi_krb5 - ubsec - padlock - safe - armv8_crypto (aarch64) - glxsb (i386) - sec (ppc) - cesa (armv7) - cryptocteon (mips64) - nlmsec (mips64) Discussed with: cem Relnotes: yes Sponsored by: Chelsio Communications Differential Revision: https://reviews.freebsd.org/D23677
2020-03-27 18:25:23 +00:00
struct g_eli_softc *sc;
crypto_freesession(wr->w_sid);
Refactor driver and consumer interfaces for OCF (in-kernel crypto). - The linked list of cryptoini structures used in session initialization is replaced with a new flat structure: struct crypto_session_params. This session includes a new mode to define how the other fields should be interpreted. Available modes include: - COMPRESS (for compression/decompression) - CIPHER (for simply encryption/decryption) - DIGEST (computing and verifying digests) - AEAD (combined auth and encryption such as AES-GCM and AES-CCM) - ETA (combined auth and encryption using encrypt-then-authenticate) Additional modes could be added in the future (e.g. if we wanted to support TLS MtE for AES-CBC in the kernel we could add a new mode for that. TLS modes might also affect how AAD is interpreted, etc.) The flat structure also includes the key lengths and algorithms as before. However, code doesn't have to walk the linked list and switch on the algorithm to determine which key is the auth key vs encryption key. The 'csp_auth_*' fields are always used for auth keys and settings and 'csp_cipher_*' for cipher. (Compression algorithms are stored in csp_cipher_alg.) - Drivers no longer register a list of supported algorithms. This doesn't quite work when you factor in modes (e.g. a driver might support both AES-CBC and SHA2-256-HMAC separately but not combined for ETA). Instead, a new 'crypto_probesession' method has been added to the kobj interface for symmteric crypto drivers. This method returns a negative value on success (similar to how device_probe works) and the crypto framework uses this value to pick the "best" driver. There are three constants for hardware (e.g. ccr), accelerated software (e.g. aesni), and plain software (cryptosoft) that give preference in that order. One effect of this is that if you request only hardware when creating a new session, you will no longer get a session using accelerated software. Another effect is that the default setting to disallow software crypto via /dev/crypto now disables accelerated software. Once a driver is chosen, 'crypto_newsession' is invoked as before. - Crypto operations are now solely described by the flat 'cryptop' structure. The linked list of descriptors has been removed. A separate enum has been added to describe the type of data buffer in use instead of using CRYPTO_F_* flags to make it easier to add more types in the future if needed (e.g. wired userspace buffers for zero-copy). It will also make it easier to re-introduce separate input and output buffers (in-kernel TLS would benefit from this). Try to make the flags related to IV handling less insane: - CRYPTO_F_IV_SEPARATE means that the IV is stored in the 'crp_iv' member of the operation structure. If this flag is not set, the IV is stored in the data buffer at the 'crp_iv_start' offset. - CRYPTO_F_IV_GENERATE means that a random IV should be generated and stored into the data buffer. This cannot be used with CRYPTO_F_IV_SEPARATE. If a consumer wants to deal with explicit vs implicit IVs, etc. it can always generate the IV however it needs and store partial IVs in the buffer and the full IV/nonce in crp_iv and set CRYPTO_F_IV_SEPARATE. The layout of the buffer is now described via fields in cryptop. crp_aad_start and crp_aad_length define the boundaries of any AAD. Previously with GCM and CCM you defined an auth crd with this range, but for ETA your auth crd had to span both the AAD and plaintext (and they had to be adjacent). crp_payload_start and crp_payload_length define the boundaries of the plaintext/ciphertext. Modes that only do a single operation (COMPRESS, CIPHER, DIGEST) should only use this region and leave the AAD region empty. If a digest is present (or should be generated), it's starting location is marked by crp_digest_start. Instead of using the CRD_F_ENCRYPT flag to determine the direction of the operation, cryptop now includes an 'op' field defining the operation to perform. For digests I've added a new VERIFY digest mode which assumes a digest is present in the input and fails the request with EBADMSG if it doesn't match the internally-computed digest. GCM and CCM already assumed this, and the new AEAD mode requires this for decryption. The new ETA mode now also requires this for decryption, so IPsec and GELI no longer do their own authentication verification. Simple DIGEST operations can also do this, though there are no in-tree consumers. To eventually support some refcounting to close races, the session cookie is now passed to crypto_getop() and clients should no longer set crp_sesssion directly. - Assymteric crypto operation structures should be allocated via crypto_getkreq() and freed via crypto_freekreq(). This permits the crypto layer to track open asym requests and close races with a driver trying to unregister while asym requests are in flight. - crypto_copyback, crypto_copydata, crypto_apply, and crypto_contiguous_subsegment now accept the 'crp' object as the first parameter instead of individual members. This makes it easier to deal with different buffer types in the future as well as separate input and output buffers. It's also simpler for driver writers to use. - bus_dmamap_load_crp() loads a DMA mapping for a crypto buffer. This understands the various types of buffers so that drivers that use DMA do not have to be aware of different buffer types. - Helper routines now exist to build an auth context for HMAC IPAD and OPAD. This reduces some duplicated work among drivers. - Key buffers are now treated as const throughout the framework and in device drivers. However, session key buffers provided when a session is created are expected to remain alive for the duration of the session. - GCM and CCM sessions now only specify a cipher algorithm and a cipher key. The redundant auth information is not needed or used. - For cryptosoft, split up the code a bit such that the 'process' callback now invokes a function pointer in the session. This function pointer is set based on the mode (in effect) though it simplifies a few edge cases that would otherwise be in the switch in 'process'. It does split up GCM vs CCM which I think is more readable even if there is some duplication. - I changed /dev/crypto to support GMAC requests using CRYPTO_AES_NIST_GMAC as an auth algorithm and updated cryptocheck to work with it. - Combined cipher and auth sessions via /dev/crypto now always use ETA mode. The COP_F_CIPHER_FIRST flag is now a no-op that is ignored. This was actually documented as being true in crypto(4) before, but the code had not implemented this before I added the CIPHER_FIRST flag. - I have not yet updated /dev/crypto to be aware of explicit modes for sessions. I will probably do that at some point in the future as well as teach it about IV/nonce and tag lengths for AEAD so we can support all of the NIST KAT tests for GCM and CCM. - I've split up the exising crypto.9 manpage into several pages of which many are written from scratch. - I have converted all drivers and consumers in the tree and verified that they compile, but I have not tested all of them. I have tested the following drivers: - cryptosoft - aesni (AES only) - blake2 - ccr and the following consumers: - cryptodev - IPsec - ktls_ocf - GELI (lightly) I have not tested the following: - ccp - aesni with sha - hifn - kgssapi_krb5 - ubsec - padlock - safe - armv8_crypto (aarch64) - glxsb (i386) - sec (ppc) - cesa (armv7) - cryptocteon (mips64) - nlmsec (mips64) Discussed with: cem Relnotes: yes Sponsored by: Chelsio Communications Differential Revision: https://reviews.freebsd.org/D23677
2020-03-27 18:25:23 +00:00
if (wr->w_first_key != NULL) {
sc = wr->w_softc;
g_eli_key_drop(sc, wr->w_first_key);
wr->w_first_key = NULL;
}
}
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
static void
g_eli_cancel(struct g_eli_softc *sc)
{
struct bio *bp;
mtx_assert(&sc->sc_queue_mtx, MA_OWNED);
while ((bp = bioq_takefirst(&sc->sc_queue)) != NULL) {
KASSERT(bp->bio_pflags == G_ELI_NEW_BIO,
("Not new bio when canceling (bp=%p).", bp));
g_io_deliver(bp, ENXIO);
}
}
static struct bio *
g_eli_takefirst(struct g_eli_softc *sc)
{
struct bio *bp;
mtx_assert(&sc->sc_queue_mtx, MA_OWNED);
if (!(sc->sc_flags & G_ELI_FLAG_SUSPEND))
return (bioq_takefirst(&sc->sc_queue));
/*
* Device suspended, so we skip new I/O requests.
*/
TAILQ_FOREACH(bp, &sc->sc_queue.queue, bio_queue) {
if (bp->bio_pflags != G_ELI_NEW_BIO)
break;
}
if (bp != NULL)
bioq_remove(&sc->sc_queue, bp);
return (bp);
}
/*
* This is the main function for kernel worker thread when we don't have
* hardware acceleration and we have to do cryptography in software.
* Dedicated thread is needed, so we don't slow down g_up/g_down GEOM
* threads with crypto work.
*/
static void
g_eli_worker(void *arg)
{
struct g_eli_softc *sc;
struct g_eli_worker *wr;
struct bio *bp;
int error;
wr = arg;
sc = wr->w_softc;
Add an EARLY_AP_STARTUP option to start APs earlier during boot. Currently, Application Processors (non-boot CPUs) are started by MD code at SI_SUB_CPU, but they are kept waiting in a "pen" until SI_SUB_SMP at which point they are released to run kernel threads. SI_SUB_SMP is one of the last SYSINIT levels, so APs don't enter the scheduler and start running threads until fairly late in the boot. This change moves SI_SUB_SMP up to just before software interrupt threads are created allowing the APs to start executing kernel threads much sooner (before any devices are probed). This allows several initialization routines that need to perform initialization on all CPUs to now perform that initialization in one step rather than having to defer the AP initialization to a second SYSINIT run at SI_SUB_SMP. It also permits all CPUs to be available for handling interrupts before any devices are probed. This last feature fixes a problem on with interrupt vector exhaustion. Specifically, in the old model all device interrupts were routed onto the boot CPU during boot. Later after the APs were released at SI_SUB_SMP, interrupts were redistributed across all CPUs. However, several drivers for multiqueue hardware allocate N interrupts per CPU in the system. In a system with many CPUs, just a few drivers doing this could exhaust the available pool of interrupt vectors on the boot CPU as each driver was allocating N * mp_ncpu vectors on the boot CPU. Now, drivers will allocate interrupts on their desired CPUs during boot meaning that only N interrupts are allocated from the boot CPU instead of N * mp_ncpu. Some other bits of code can also be simplified as smp_started is now true much earlier and will now always be true for these bits of code. This removes the need to treat the single-CPU boot environment as a special case. As a transition aid, the new behavior is available under a new kernel option (EARLY_AP_STARTUP). This will allow the option to be turned off if need be during initial testing. I plan to enable this on x86 by default in a followup commit in the next few days and to have all platforms moved over before 11.0. Once the transition is complete, the option will be removed along with the !EARLY_AP_STARTUP code. These changes have only been tested on x86. Other platform maintainers are encouraged to port their architectures over as well. The main things to check for are any uses of smp_started in MD code that can be simplified and SI_SUB_SMP SYSINITs in MD code that can be removed in the EARLY_AP_STARTUP case (e.g. the interrupt shuffling). PR: kern/199321 Reviewed by: markj, gnn, kib Sponsored by: Netflix
2016-05-14 18:22:52 +00:00
#ifdef EARLY_AP_STARTUP
MPASS(!sc->sc_cpubind || smp_started);
#elif defined(SMP)
/* Before sched_bind() to a CPU, wait for all CPUs to go on-line. */
if (sc->sc_cpubind) {
while (!smp_started)
tsleep(wr, 0, "geli:smp", hz / 4);
}
#endif
thread_lock(curthread);
sched_prio(curthread, PUSER);
if (sc->sc_cpubind)
sched_bind(curthread, wr->w_number % mp_ncpus);
thread_unlock(curthread);
2006-02-01 12:06:01 +00:00
G_ELI_DEBUG(1, "Thread %s started.", curthread->td_proc->p_comm);
for (;;) {
mtx_lock(&sc->sc_queue_mtx);
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
again:
bp = g_eli_takefirst(sc);
if (bp == NULL) {
if (sc->sc_flags & G_ELI_FLAG_DESTROY) {
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
g_eli_cancel(sc);
LIST_REMOVE(wr, w_next);
g_eli_freesession(wr);
free(wr, M_ELI);
G_ELI_DEBUG(1, "Thread %s exiting.",
curthread->td_proc->p_comm);
wakeup(&sc->sc_workers);
mtx_unlock(&sc->sc_queue_mtx);
kproc_exit(0);
}
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
while (sc->sc_flags & G_ELI_FLAG_SUSPEND) {
if (sc->sc_inflight > 0) {
G_ELI_DEBUG(0, "inflight=%d",
sc->sc_inflight);
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
/*
* We still have inflight BIOs, so
* sleep and retry.
*/
msleep(sc, &sc->sc_queue_mtx, PRIBIO,
"geli:inf", hz / 5);
goto again;
}
/*
* Suspend requested, mark the worker as
* suspended and go to sleep.
*/
if (wr->w_active) {
g_eli_freesession(wr);
wr->w_active = FALSE;
}
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
wakeup(&sc->sc_workers);
msleep(sc, &sc->sc_queue_mtx, PRIBIO,
"geli:suspend", 0);
if (!wr->w_active &&
!(sc->sc_flags & G_ELI_FLAG_SUSPEND)) {
error = g_eli_newsession(wr);
KASSERT(error == 0,
("g_eli_newsession() failed on resume (error=%d)",
error));
wr->w_active = TRUE;
}
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
goto again;
}
msleep(sc, &sc->sc_queue_mtx, PDROP, "geli:w", 0);
continue;
}
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
if (bp->bio_pflags == G_ELI_NEW_BIO)
atomic_add_int(&sc->sc_inflight, 1);
mtx_unlock(&sc->sc_queue_mtx);
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
if (bp->bio_pflags == G_ELI_NEW_BIO) {
bp->bio_pflags = 0;
if (sc->sc_flags & G_ELI_FLAG_AUTH) {
if (bp->bio_cmd == BIO_READ)
g_eli_auth_read(sc, bp);
else
g_eli_auth_run(wr, bp);
} else {
if (bp->bio_cmd == BIO_READ)
g_eli_crypto_read(sc, bp, 1);
else
g_eli_crypto_run(wr, bp);
}
} else {
if (sc->sc_flags & G_ELI_FLAG_AUTH)
g_eli_auth_run(wr, bp);
else
g_eli_crypto_run(wr, bp);
}
}
}
static int
g_eli_read_metadata_offset(struct g_class *mp, struct g_provider *pp,
off_t offset, struct g_eli_metadata *md)
{
struct g_geom *gp;
struct g_consumer *cp;
u_char *buf = NULL;
int error;
g_topology_assert();
gp = g_new_geomf(mp, "eli:taste");
gp->start = g_eli_start;
gp->access = g_std_access;
/*
* g_eli_read_metadata() is always called from the event thread.
* Our geom is created and destroyed in the same event, so there
* could be no orphan nor spoil event in the meantime.
*/
gp->orphan = g_eli_orphan_spoil_assert;
gp->spoiled = g_eli_orphan_spoil_assert;
cp = g_new_consumer(gp);
cp->flags |= G_CF_DIRECT_SEND | G_CF_DIRECT_RECEIVE;
error = g_attach(cp, pp);
if (error != 0)
goto end;
error = g_access(cp, 1, 0, 0);
if (error != 0)
goto end;
g_topology_unlock();
buf = g_read_data(cp, offset, pp->sectorsize, &error);
g_topology_lock();
if (buf == NULL)
goto end;
error = eli_metadata_decode(buf, md);
if (error != 0)
goto end;
/* Metadata was read and decoded successfully. */
end:
if (buf != NULL)
g_free(buf);
if (cp->provider != NULL) {
if (cp->acr == 1)
g_access(cp, -1, 0, 0);
g_detach(cp);
}
g_destroy_consumer(cp);
g_destroy_geom(gp);
return (error);
}
int
g_eli_read_metadata(struct g_class *mp, struct g_provider *pp,
struct g_eli_metadata *md)
{
return (g_eli_read_metadata_offset(mp, pp,
pp->mediasize - pp->sectorsize, md));
}
/*
* The function is called when we had last close on provider and user requested
* to close it when this situation occur.
*/
static void
g_eli_last_close(void *arg, int flags __unused)
{
struct g_geom *gp;
char gpname[64];
int error;
g_topology_assert();
gp = arg;
strlcpy(gpname, gp->name, sizeof(gpname));
error = g_eli_destroy(gp->softc, TRUE);
KASSERT(error == 0, ("Cannot detach %s on last close (error=%d).",
gpname, error));
G_ELI_DEBUG(0, "Detached %s on last close.", gpname);
}
int
g_eli_access(struct g_provider *pp, int dr, int dw, int de)
{
struct g_eli_softc *sc;
struct g_geom *gp;
gp = pp->geom;
sc = gp->softc;
if (dw > 0) {
if (sc->sc_flags & G_ELI_FLAG_RO) {
/* Deny write attempts. */
return (EROFS);
}
/* Someone is opening us for write, we need to remember that. */
sc->sc_flags |= G_ELI_FLAG_WOPEN;
return (0);
}
/* Is this the last close? */
if (pp->acr + dr > 0 || pp->acw + dw > 0 || pp->ace + de > 0)
return (0);
/*
* Automatically detach on last close if requested.
*/
if ((sc->sc_flags & G_ELI_FLAG_RW_DETACH) ||
(sc->sc_flags & G_ELI_FLAG_WOPEN)) {
g_post_event(g_eli_last_close, gp, M_WAITOK, NULL);
}
return (0);
}
static int
g_eli_cpu_is_disabled(int cpu)
{
#ifdef SMP
Commit the support for removing cpumask_t and replacing it directly with cpuset_t objects. That is going to offer the underlying support for a simple bump of MAXCPU and then support for number of cpus > 32 (as it is today). Right now, cpumask_t is an int, 32 bits on all our supported architecture. cpumask_t on the other side is implemented as an array of longs, and easilly extendible by definition. The architectures touched by this commit are the following: - amd64 - i386 - pc98 - arm - ia64 - XEN while the others are still missing. Userland is believed to be fully converted with the changes contained here. Some technical notes: - This commit may be considered an ABI nop for all the architectures different from amd64 and ia64 (and sparc64 in the future) - per-cpu members, which are now converted to cpuset_t, needs to be accessed avoiding migration, because the size of cpuset_t should be considered unknown - size of cpuset_t objects is different from kernel and userland (this is primirally done in order to leave some more space in userland to cope with KBI extensions). If you need to access kernel cpuset_t from the userland please refer to example in this patch on how to do that correctly (kgdb may be a good source, for example). - Support for other architectures is going to be added soon - Only MAXCPU for amd64 is bumped now The patch has been tested by sbruno and Nicholas Esborn on opteron 4 x 12 pack CPUs. More testing on big SMP is expected to came soon. pluknet tested the patch with his 8-ways on both amd64 and i386. Tested by: pluknet, sbruno, gianni, Nicholas Esborn Reviewed by: jeff, jhb, sbruno
2011-05-05 14:39:14 +00:00
return (CPU_ISSET(cpu, &hlt_cpus_mask));
#else
return (0);
#endif
}
struct g_geom *
g_eli_create(struct gctl_req *req, struct g_class *mp, struct g_provider *bpp,
const struct g_eli_metadata *md, const u_char *mkey, int nkey)
{
struct g_eli_softc *sc;
struct g_eli_worker *wr;
struct g_geom *gp;
struct g_provider *pp;
struct g_consumer *cp;
struct g_geom_alias *gap;
u_int i, threads;
int dcw, error;
G_ELI_DEBUG(1, "Creating device %s%s.", bpp->name, G_ELI_SUFFIX);
KASSERT(eli_metadata_crypto_supported(md),
("%s: unsupported crypto for %s", __func__, bpp->name));
gp = g_new_geomf(mp, "%s%s", bpp->name, G_ELI_SUFFIX);
sc = malloc(sizeof(*sc), M_ELI, M_WAITOK | M_ZERO);
gp->start = g_eli_start;
/*
* Spoiling can happen even though we have the provider open
* exclusively, e.g. through media change events.
*/
gp->spoiled = g_eli_orphan;
gp->orphan = g_eli_orphan;
gp->resize = g_eli_resize;
gp->dumpconf = g_eli_dumpconf;
/*
* If detach-on-last-close feature is not enabled and we don't operate
* on read-only provider, we can simply use g_std_access().
*/
if (md->md_flags & (G_ELI_FLAG_WO_DETACH | G_ELI_FLAG_RO))
gp->access = g_eli_access;
else
gp->access = g_std_access;
eli_metadata_softc(sc, md, bpp->sectorsize, bpp->mediasize);
sc->sc_nkey = nkey;
gp->softc = sc;
sc->sc_geom = gp;
bioq_init(&sc->sc_queue);
mtx_init(&sc->sc_queue_mtx, "geli:queue", NULL, MTX_DEF);
mtx_init(&sc->sc_ekeys_lock, "geli:ekeys", NULL, MTX_DEF);
pp = NULL;
cp = g_new_consumer(gp);
cp->flags |= G_CF_DIRECT_SEND | G_CF_DIRECT_RECEIVE;
error = g_attach(cp, bpp);
if (error != 0) {
if (req != NULL) {
gctl_error(req, "Cannot attach to %s (error=%d).",
bpp->name, error);
} else {
G_ELI_DEBUG(1, "Cannot attach to %s (error=%d).",
bpp->name, error);
}
goto failed;
}
/*
* Keep provider open all the time, so we can run critical tasks,
* like Master Keys deletion, without wondering if we can open
* provider or not.
* We don't open provider for writing only when user requested read-only
* access.
*/
dcw = (sc->sc_flags & G_ELI_FLAG_RO) ? 0 : 1;
error = g_access(cp, 1, dcw, 1);
if (error != 0) {
if (req != NULL) {
gctl_error(req, "Cannot access %s (error=%d).",
bpp->name, error);
} else {
G_ELI_DEBUG(1, "Cannot access %s (error=%d).",
bpp->name, error);
}
goto failed;
}
/*
* Remember the keys in our softc structure.
*/
g_eli_mkey_propagate(sc, mkey);
LIST_INIT(&sc->sc_workers);
threads = g_eli_threads;
if (threads == 0)
threads = mp_ncpus;
sc->sc_cpubind = (mp_ncpus > 1 && threads == mp_ncpus);
for (i = 0; i < threads; i++) {
if (g_eli_cpu_is_disabled(i)) {
G_ELI_DEBUG(1, "%s: CPU %u disabled, skipping.",
2006-11-02 09:01:34 +00:00
bpp->name, i);
continue;
}
wr = malloc(sizeof(*wr), M_ELI, M_WAITOK | M_ZERO);
wr->w_softc = sc;
wr->w_number = i;
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
wr->w_active = TRUE;
error = g_eli_newsession(wr);
if (error != 0) {
free(wr, M_ELI);
if (req != NULL) {
gctl_error(req, "Cannot set up crypto session "
"for %s (error=%d).", bpp->name, error);
} else {
G_ELI_DEBUG(1, "Cannot set up crypto session "
"for %s (error=%d).", bpp->name, error);
}
goto failed;
}
error = kproc_create(g_eli_worker, wr, &wr->w_proc, 0, 0,
"g_eli[%u] %s", i, bpp->name);
if (error != 0) {
g_eli_freesession(wr);
free(wr, M_ELI);
if (req != NULL) {
gctl_error(req, "Cannot create kernel thread "
"for %s (error=%d).", bpp->name, error);
} else {
G_ELI_DEBUG(1, "Cannot create kernel thread "
"for %s (error=%d).", bpp->name, error);
}
goto failed;
}
LIST_INSERT_HEAD(&sc->sc_workers, wr, w_next);
}
/*
* Create decrypted provider.
*/
pp = g_new_providerf(gp, "%s%s", bpp->name, G_ELI_SUFFIX);
pp->flags |= G_PF_DIRECT_SEND | G_PF_DIRECT_RECEIVE;
pp->mediasize = sc->sc_mediasize;
pp->sectorsize = sc->sc_sectorsize;
LIST_FOREACH(gap, &bpp->aliases, ga_next)
g_provider_add_alias(pp, "%s%s", gap->ga_alias, G_ELI_SUFFIX);
g_error_provider(pp, 0);
G_ELI_DEBUG(0, "Device %s created.", pp->name);
G_ELI_DEBUG(0, "Encryption: %s %u", g_eli_algo2str(sc->sc_ealgo),
sc->sc_ekeylen);
if (sc->sc_flags & G_ELI_FLAG_AUTH)
G_ELI_DEBUG(0, " Integrity: %s", g_eli_algo2str(sc->sc_aalgo));
G_ELI_DEBUG(0, " Crypto: %s",
sc->sc_crypto == G_ELI_CRYPTO_SW_ACCEL ? "accelerated software" :
sc->sc_crypto == G_ELI_CRYPTO_SW ? "software" : "hardware");
return (gp);
failed:
mtx_lock(&sc->sc_queue_mtx);
sc->sc_flags |= G_ELI_FLAG_DESTROY;
wakeup(sc);
/*
* Wait for kernel threads self destruction.
*/
while (!LIST_EMPTY(&sc->sc_workers)) {
msleep(&sc->sc_workers, &sc->sc_queue_mtx, PRIBIO,
"geli:destroy", 0);
}
mtx_destroy(&sc->sc_queue_mtx);
if (cp->provider != NULL) {
if (cp->acr == 1)
g_access(cp, -1, -dcw, -1);
g_detach(cp);
}
g_destroy_consumer(cp);
g_destroy_geom(gp);
g_eli_key_destroy(sc);
zfree(sc, M_ELI);
return (NULL);
}
int
g_eli_destroy(struct g_eli_softc *sc, boolean_t force)
{
struct g_geom *gp;
struct g_provider *pp;
g_topology_assert();
if (sc == NULL)
return (ENXIO);
gp = sc->sc_geom;
pp = LIST_FIRST(&gp->provider);
if (pp != NULL && (pp->acr != 0 || pp->acw != 0 || pp->ace != 0)) {
if (force) {
G_ELI_DEBUG(1, "Device %s is still open, so it "
"cannot be definitely removed.", pp->name);
sc->sc_flags |= G_ELI_FLAG_RW_DETACH;
gp->access = g_eli_access;
g_wither_provider(pp, ENXIO);
return (EBUSY);
} else {
G_ELI_DEBUG(1,
"Device %s is still open (r%dw%de%d).", pp->name,
pp->acr, pp->acw, pp->ace);
return (EBUSY);
}
}
mtx_lock(&sc->sc_queue_mtx);
sc->sc_flags |= G_ELI_FLAG_DESTROY;
wakeup(sc);
while (!LIST_EMPTY(&sc->sc_workers)) {
msleep(&sc->sc_workers, &sc->sc_queue_mtx, PRIBIO,
"geli:destroy", 0);
}
mtx_destroy(&sc->sc_queue_mtx);
gp->softc = NULL;
g_eli_key_destroy(sc);
zfree(sc, M_ELI);
G_ELI_DEBUG(0, "Device %s destroyed.", gp->name);
g_wither_geom_close(gp, ENXIO);
return (0);
}
static int
g_eli_destroy_geom(struct gctl_req *req __unused,
struct g_class *mp __unused, struct g_geom *gp)
{
struct g_eli_softc *sc;
sc = gp->softc;
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
return (g_eli_destroy(sc, FALSE));
}
static int
g_eli_keyfiles_load(struct hmac_ctx *ctx, const char *provider)
{
u_char *keyfile, *data;
char *file, name[64];
size_t size;
int i;
for (i = 0; ; i++) {
snprintf(name, sizeof(name), "%s:geli_keyfile%d", provider, i);
keyfile = preload_search_by_type(name);
if (keyfile == NULL && i == 0) {
/*
* If there is only one keyfile, allow simpler name.
*/
snprintf(name, sizeof(name), "%s:geli_keyfile", provider);
keyfile = preload_search_by_type(name);
}
if (keyfile == NULL)
return (i); /* Return number of loaded keyfiles. */
data = preload_fetch_addr(keyfile);
if (data == NULL) {
G_ELI_DEBUG(0, "Cannot find key file data for %s.",
name);
return (0);
}
size = preload_fetch_size(keyfile);
if (size == 0) {
G_ELI_DEBUG(0, "Cannot find key file size for %s.",
name);
return (0);
}
file = preload_search_info(keyfile, MODINFO_NAME);
if (file == NULL) {
G_ELI_DEBUG(0, "Cannot find key file name for %s.",
name);
return (0);
}
G_ELI_DEBUG(1, "Loaded keyfile %s for %s (type: %s).", file,
provider, name);
g_eli_crypto_hmac_update(ctx, data, size);
}
}
static void
g_eli_keyfiles_clear(const char *provider)
{
u_char *keyfile, *data;
char name[64];
size_t size;
int i;
for (i = 0; ; i++) {
snprintf(name, sizeof(name), "%s:geli_keyfile%d", provider, i);
keyfile = preload_search_by_type(name);
if (keyfile == NULL)
return;
data = preload_fetch_addr(keyfile);
size = preload_fetch_size(keyfile);
if (data != NULL && size != 0)
explicit_bzero(data, size);
}
}
/*
* Tasting is only made on boot.
* We detect providers which should be attached before root is mounted.
*/
static struct g_geom *
g_eli_taste(struct g_class *mp, struct g_provider *pp, int flags __unused)
{
struct g_eli_metadata md;
struct g_geom *gp;
struct hmac_ctx ctx;
char passphrase[256];
u_char key[G_ELI_USERKEYLEN], mkey[G_ELI_DATAIVKEYLEN];
u_int i, nkey, nkeyfiles, tries, showpass;
int error;
struct keybuf *keybuf;
g_trace(G_T_TOPOLOGY, "%s(%s, %s)", __func__, mp->name, pp->name);
g_topology_assert();
2007-04-08 23:54:23 +00:00
if (root_mounted() || g_eli_tries == 0)
return (NULL);
G_ELI_DEBUG(3, "Tasting %s.", pp->name);
error = g_eli_read_metadata(mp, pp, &md);
if (error != 0)
return (NULL);
gp = NULL;
if (strcmp(md.md_magic, G_ELI_MAGIC) != 0)
return (NULL);
if (md.md_version > G_ELI_VERSION) {
printf("geom_eli.ko module is too old to handle %s.\n",
pp->name);
return (NULL);
}
if (md.md_provsize != pp->mediasize)
return (NULL);
/* Should we attach it on boot? */
if (!(md.md_flags & G_ELI_FLAG_BOOT) &&
!(md.md_flags & G_ELI_FLAG_GELIBOOT))
return (NULL);
if (md.md_keys == 0x00) {
G_ELI_DEBUG(0, "No valid keys on %s.", pp->name);
return (NULL);
}
if (!eli_metadata_crypto_supported(&md)) {
G_ELI_DEBUG(0, "%s uses invalid or unsupported algorithms\n",
pp->name);
return (NULL);
}
if (md.md_iterations == -1) {
/* If there is no passphrase, we try only once. */
tries = 1;
} else {
/* Ask for the passphrase no more than g_eli_tries times. */
tries = g_eli_tries;
}
if ((keybuf = get_keybuf()) != NULL) {
/* Scan the key buffer, try all GELI keys. */
for (i = 0; i < keybuf->kb_nents; i++) {
if (keybuf->kb_ents[i].ke_type == KEYBUF_TYPE_GELI) {
memcpy(key, keybuf->kb_ents[i].ke_data,
sizeof(key));
if (g_eli_mkey_decrypt_any(&md, key,
mkey, &nkey) == 0 ) {
explicit_bzero(key, sizeof(key));
goto have_key;
}
}
}
}
for (i = 0; i <= tries; i++) {
g_eli_crypto_hmac_init(&ctx, NULL, 0);
/*
* Load all key files.
*/
nkeyfiles = g_eli_keyfiles_load(&ctx, pp->name);
if (nkeyfiles == 0 && md.md_iterations == -1) {
/*
* No key files and no passphrase, something is
* definitely wrong here.
* geli(8) doesn't allow for such situation, so assume
* that there was really no passphrase and in that case
* key files are no properly defined in loader.conf.
*/
G_ELI_DEBUG(0,
"Found no key files in loader.conf for %s.",
pp->name);
return (NULL);
}
/* Ask for the passphrase if defined. */
if (md.md_iterations >= 0) {
/* Try first with cached passphrase. */
if (i == 0) {
if (!g_eli_boot_passcache)
continue;
memcpy(passphrase, cached_passphrase,
sizeof(passphrase));
} else {
printf("Enter passphrase for %s: ", pp->name);
showpass = g_eli_visible_passphrase;
if ((md.md_flags & G_ELI_FLAG_GELIDISPLAYPASS) != 0)
showpass = GETS_ECHOPASS;
cngets(passphrase, sizeof(passphrase),
showpass);
memcpy(cached_passphrase, passphrase,
sizeof(passphrase));
}
}
/*
* Prepare Derived-Key from the user passphrase.
*/
if (md.md_iterations == 0) {
g_eli_crypto_hmac_update(&ctx, md.md_salt,
sizeof(md.md_salt));
g_eli_crypto_hmac_update(&ctx, passphrase,
strlen(passphrase));
explicit_bzero(passphrase, sizeof(passphrase));
} else if (md.md_iterations > 0) {
u_char dkey[G_ELI_USERKEYLEN];
pkcs5v2_genkey(dkey, sizeof(dkey), md.md_salt,
sizeof(md.md_salt), passphrase, md.md_iterations);
explicit_bzero(passphrase, sizeof(passphrase));
g_eli_crypto_hmac_update(&ctx, dkey, sizeof(dkey));
explicit_bzero(dkey, sizeof(dkey));
}
g_eli_crypto_hmac_final(&ctx, key, 0);
/*
* Decrypt Master-Key.
*/
error = g_eli_mkey_decrypt_any(&md, key, mkey, &nkey);
explicit_bzero(key, sizeof(key));
if (error == -1) {
if (i == tries) {
G_ELI_DEBUG(0,
"Wrong key for %s. No tries left.",
pp->name);
g_eli_keyfiles_clear(pp->name);
return (NULL);
}
if (i > 0) {
G_ELI_DEBUG(0,
"Wrong key for %s. Tries left: %u.",
pp->name, tries - i);
}
/* Try again. */
continue;
} else if (error > 0) {
G_ELI_DEBUG(0,
"Cannot decrypt Master Key for %s (error=%d).",
pp->name, error);
g_eli_keyfiles_clear(pp->name);
return (NULL);
}
g_eli_keyfiles_clear(pp->name);
G_ELI_DEBUG(1, "Using Master Key %u for %s.", nkey, pp->name);
break;
}
have_key:
/*
* We have correct key, let's attach provider.
*/
gp = g_eli_create(NULL, mp, pp, &md, mkey, nkey);
explicit_bzero(mkey, sizeof(mkey));
explicit_bzero(&md, sizeof(md));
if (gp == NULL) {
G_ELI_DEBUG(0, "Cannot create device %s%s.", pp->name,
G_ELI_SUFFIX);
return (NULL);
}
return (gp);
}
static void
g_eli_dumpconf(struct sbuf *sb, const char *indent, struct g_geom *gp,
struct g_consumer *cp, struct g_provider *pp)
{
struct g_eli_softc *sc;
g_topology_assert();
sc = gp->softc;
if (sc == NULL)
return;
if (pp != NULL || cp != NULL)
return; /* Nothing here. */
sbuf_printf(sb, "%s<KeysTotal>%ju</KeysTotal>\n", indent,
(uintmax_t)sc->sc_ekeys_total);
sbuf_printf(sb, "%s<KeysAllocated>%ju</KeysAllocated>\n", indent,
(uintmax_t)sc->sc_ekeys_allocated);
sbuf_printf(sb, "%s<Flags>", indent);
if (sc->sc_flags == 0)
sbuf_cat(sb, "NONE");
else {
int first = 1;
#define ADD_FLAG(flag, name) do { \
if (sc->sc_flags & (flag)) { \
if (!first) \
sbuf_cat(sb, ", "); \
else \
first = 0; \
sbuf_cat(sb, name); \
} \
} while (0)
Bring in geli suspend/resume functionality (finally). Before this change if you wanted to suspend your laptop and be sure that your encryption keys are safe, you had to stop all processes that use file system stored on encrypted device, unmount the file system and detach geli provider. This isn't very handy. If you are a lucky user of a laptop where suspend/resume actually works with FreeBSD (I'm not!) you most likely want to suspend your laptop, because you don't want to start everything over again when you turn your laptop back on. And this is where geli suspend/resume steps in. When you execute: # geli suspend -a geli will wait for all in-flight I/O requests, suspend new I/O requests, remove all geli sensitive data from the kernel memory (like encryption keys) and will wait for either 'geli resume' or 'geli detach'. Now with no keys in memory you can suspend your laptop without stopping any processes or unmounting any file systems. When you resume your laptop you have to resume geli devices using 'geli resume' command. You need to provide your passphrase, etc. again so the keys can be restored and suspended I/O requests released. Of course you need to remember that 'geli suspend' won't clear file system cache and other places where data from your geli-encrypted file system might be present. But to get rid of those stopping processes and unmounting file system won't help either - you have to turn your laptop off. Be warned. Also note, that suspending geli device which contains file system with geli utility (or anything used by 'geli resume') is not very good idea, as you won't be able to resume it - when you execute geli(8), the kernel will try to read it and this read I/O request will be suspended.
2010-10-20 20:50:55 +00:00
ADD_FLAG(G_ELI_FLAG_SUSPEND, "SUSPEND");
ADD_FLAG(G_ELI_FLAG_SINGLE_KEY, "SINGLE-KEY");
ADD_FLAG(G_ELI_FLAG_NATIVE_BYTE_ORDER, "NATIVE-BYTE-ORDER");
ADD_FLAG(G_ELI_FLAG_ONETIME, "ONETIME");
ADD_FLAG(G_ELI_FLAG_BOOT, "BOOT");
ADD_FLAG(G_ELI_FLAG_WO_DETACH, "W-DETACH");
ADD_FLAG(G_ELI_FLAG_RW_DETACH, "RW-DETACH");
ADD_FLAG(G_ELI_FLAG_AUTH, "AUTH");
ADD_FLAG(G_ELI_FLAG_WOPEN, "W-OPEN");
ADD_FLAG(G_ELI_FLAG_DESTROY, "DESTROY");
ADD_FLAG(G_ELI_FLAG_RO, "READ-ONLY");
ADD_FLAG(G_ELI_FLAG_NODELETE, "NODELETE");
ADD_FLAG(G_ELI_FLAG_GELIBOOT, "GELIBOOT");
ADD_FLAG(G_ELI_FLAG_GELIDISPLAYPASS, "GELIDISPLAYPASS");
ADD_FLAG(G_ELI_FLAG_AUTORESIZE, "AUTORESIZE");
#undef ADD_FLAG
}
sbuf_cat(sb, "</Flags>\n");
if (!(sc->sc_flags & G_ELI_FLAG_ONETIME)) {
sbuf_printf(sb, "%s<UsedKey>%u</UsedKey>\n", indent,
sc->sc_nkey);
}
sbuf_printf(sb, "%s<Version>%u</Version>\n", indent, sc->sc_version);
sbuf_printf(sb, "%s<Crypto>", indent);
switch (sc->sc_crypto) {
case G_ELI_CRYPTO_HW:
sbuf_cat(sb, "hardware");
break;
case G_ELI_CRYPTO_SW:
sbuf_cat(sb, "software");
break;
case G_ELI_CRYPTO_SW_ACCEL:
sbuf_cat(sb, "accelerated software");
break;
default:
sbuf_cat(sb, "UNKNOWN");
break;
}
sbuf_cat(sb, "</Crypto>\n");
if (sc->sc_flags & G_ELI_FLAG_AUTH) {
sbuf_printf(sb,
"%s<AuthenticationAlgorithm>%s</AuthenticationAlgorithm>\n",
indent, g_eli_algo2str(sc->sc_aalgo));
}
sbuf_printf(sb, "%s<KeyLength>%u</KeyLength>\n", indent,
sc->sc_ekeylen);
sbuf_printf(sb, "%s<EncryptionAlgorithm>%s</EncryptionAlgorithm>\n",
indent, g_eli_algo2str(sc->sc_ealgo));
sbuf_printf(sb, "%s<State>%s</State>\n", indent,
(sc->sc_flags & G_ELI_FLAG_SUSPEND) ? "SUSPENDED" : "ACTIVE");
}
static void
g_eli_shutdown_pre_sync(void *arg, int howto)
{
struct g_class *mp;
struct g_geom *gp, *gp2;
struct g_provider *pp;
struct g_eli_softc *sc;
int error;
mp = arg;
g_topology_lock();
LIST_FOREACH_SAFE(gp, &mp->geom, geom, gp2) {
sc = gp->softc;
if (sc == NULL)
continue;
pp = LIST_FIRST(&gp->provider);
KASSERT(pp != NULL, ("No provider? gp=%p (%s)", gp, gp->name));
if (pp->acr != 0 || pp->acw != 0 || pp->ace != 0 ||
SCHEDULER_STOPPED())
{
sc->sc_flags |= G_ELI_FLAG_RW_DETACH;
gp->access = g_eli_access;
} else {
error = g_eli_destroy(sc, TRUE);
}
}
g_topology_unlock();
}
static void
g_eli_init(struct g_class *mp)
{
g_eli_pre_sync = EVENTHANDLER_REGISTER(shutdown_pre_sync,
g_eli_shutdown_pre_sync, mp, SHUTDOWN_PRI_FIRST);
if (g_eli_pre_sync == NULL)
G_ELI_DEBUG(0, "Warning! Cannot register shutdown event.");
}
static void
g_eli_fini(struct g_class *mp)
{
if (g_eli_pre_sync != NULL)
EVENTHANDLER_DEREGISTER(shutdown_pre_sync, g_eli_pre_sync);
}
DECLARE_GEOM_CLASS(g_eli_class, g_eli);
MODULE_DEPEND(g_eli, crypto, 1, 1, 1);
MODULE_VERSION(geom_eli, 0);