d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
ddf8a6680e
freebsd-dev/sys/geom/geom_io.c
Kenneth D. Merry a9934668aa Add asynchronous command support to the pass(4) driver, and the new 
camdd(8) utility.

CCBs may be queued to the driver via the new CAMIOQUEUE ioctl, and
completed CCBs may be retrieved via the CAMIOGET ioctl.  User
processes can use poll(2) or kevent(2) to get notification when
I/O has completed.

While the existing CAMIOCOMMAND blocking ioctl interface only
supports user virtual data pointers in a CCB (generally only
one per CCB), the new CAMIOQUEUE ioctl supports user virtual and
physical address pointers, as well as user virtual and physical
scatter/gather lists.  This allows user applications to have more
flexibility in their data handling operations.

Kernel memory for data transferred via the queued interface is
allocated from the zone allocator in MAXPHYS sized chunks, and user
data is copied in and out.  This is likely faster than the
vmapbuf()/vunmapbuf() method used by the CAMIOCOMMAND ioctl in
configurations with many processors (there are more TLB shootdowns
caused by the mapping/unmapping operation) but may not be as fast
as running with unmapped I/O.

The new memory handling model for user requests also allows
applications to send CCBs with request sizes that are larger than
MAXPHYS.  The pass(4) driver now limits queued requests to the I/O
size listed by the SIM driver in the maxio field in the Path
Inquiry (XPT_PATH_INQ) CCB.

There are some things things would be good to add:

1. Come up with a way to do unmapped I/O on multiple buffers.
   Currently the unmapped I/O interface operates on a struct bio,
   which includes only one address and length.  It would be nice
   to be able to send an unmapped scatter/gather list down to
   busdma.  This would allow eliminating the copy we currently do
   for data.

2. Add an ioctl to list currently outstanding CCBs in the various
   queues.

3. Add an ioctl to cancel a request, or use the XPT_ABORT CCB to do
   that.

4. Test physical address support.  Virtual pointers and scatter
   gather lists have been tested, but I have not yet tested
   physical addresses or scatter/gather lists.

5. Investigate multiple queue support.  At the moment there is one
   queue of commands per pass(4) device.  If multiple processes
   open the device, they will submit I/O into the same queue and
   get events for the same completions.  This is probably the right
   model for most applications, but it is something that could be
   changed later on.

Also, add a new utility, camdd(8) that uses the asynchronous pass(4)
driver interface.

This utility is intended to be a basic data transfer/copy utility,
a simple benchmark utility, and an example of how to use the
asynchronous pass(4) interface.

It can copy data to and from pass(4) devices using any target queue
depth, starting offset and blocksize for the input and ouptut devices.
It currently only supports SCSI devices, but could be easily extended
to support ATA devices.

It can also copy data to and from regular files, block devices, tape
devices, pipes, stdin, and stdout.  It does not support queueing
multiple commands to any of those targets, since it uses the standard
read(2)/write(2)/writev(2)/readv(2) system calls.

The I/O is done by two threads, one for the reader and one for the
writer.  The reader thread sends completed read requests to the
writer thread in strictly sequential order, even if they complete
out of order.  That could be modified later on for random I/O patterns
or slightly out of order I/O.

camdd(8) uses kqueue(2)/kevent(2) to get I/O completion events from
the pass(4) driver and also to send request notifications internally.

For pass(4) devcies, camdd(8) uses a single buffer (CAM_DATA_VADDR)
per CAM CCB on the reading side, and a scatter/gather list
(CAM_DATA_SG) on the writing side.  In addition to testing both
interfaces, this makes any potential reblocking of I/O easier.  No
data is copied between the reader and the writer, but rather the
reader's buffers are split into multiple I/O requests or combined
into a single I/O request depending on the input and output blocksize.

For the file I/O path, camdd(8) also uses a single buffer (read(2),
write(2), pread(2) or pwrite(2)) on reads, and a scatter/gather list
(readv(2), writev(2), preadv(2), pwritev(2)) on writes.

Things that would be nice to do for camdd(8) eventually:

1.  Add support for I/O pattern generation.  Patterns like all
    zeros, all ones, LBA-based patterns, random patterns, etc. Right
    Now you can always use /dev/zero, /dev/random, etc.

2.  Add support for a "sink" mode, so we do only reads with no
    writes.  Right now, you can use /dev/null.

3.  Add support for automatic queue depth probing, so that we can
    figure out the right queue depth on the input and output side
    for maximum throughput.  At the moment it defaults to 6.

4.  Add support for SATA device passthrough I/O.

5.  Add support for random LBAs and/or lengths on the input and
    output sides.

6.  Track average per-I/O latency and busy time.  The busy time
    and latency could also feed in to the automatic queue depth
    determination.

sys/cam/scsi/scsi_pass.h:
	Define two new ioctls, CAMIOQUEUE and CAMIOGET, that queue
	and fetch asynchronous CAM CCBs respectively.

	Although these ioctls do not have a declared argument, they
	both take a union ccb pointer.  If we declare a size here,
	the ioctl code in sys/kern/sys_generic.c will malloc and free
	a buffer for either the CCB or the CCB pointer (depending on
	how it is declared).  Since we have to keep a copy of the
	CCB (which is fairly large) anyway, having the ioctl malloc
	and free a CCB for each call is wasteful.

sys/cam/scsi/scsi_pass.c:
	Add asynchronous CCB support.

	Add two new ioctls, CAMIOQUEUE and CAMIOGET.

	CAMIOQUEUE adds a CCB to the incoming queue.  The CCB is
	executed immediately (and moved to the active queue) if it
	is an immediate CCB, but otherwise it will be executed
	in passstart() when a CCB is available from the transport layer.

	When CCBs are completed (because they are immediate or
	passdone() if they are queued), they are put on the done
	queue.

	If we get the final close on the device before all pending
	I/O is complete, all active I/O is moved to the abandoned
	queue and we increment the peripheral reference count so
	that the peripheral driver instance doesn't go away before
	all pending I/O is done.

	The new passcreatezone() function is called on the first
	call to the CAMIOQUEUE ioctl on a given device to allocate
	the UMA zones for I/O requests and S/G list buffers.  This
	may be good to move off to a taskqueue at some point.
	The new passmemsetup() function allocates memory and
	scatter/gather lists to hold the user's data, and copies
	in any data that needs to be written.  For virtual pointers
	(CAM_DATA_VADDR), the kernel buffer is malloced from the
	new pass(4) driver malloc bucket.  For virtual
	scatter/gather lists (CAM_DATA_SG), buffers are allocated
	from a new per-pass(9) UMA zone in MAXPHYS-sized chunks.
	Physical pointers are passed in unchanged.  We have support
	for up to 16 scatter/gather segments (for the user and
	kernel S/G lists) in the default struct pass_io_req, so
	requests with longer S/G lists require an extra kernel malloc.

	The new passcopysglist() function copies a user scatter/gather
	list to a kernel scatter/gather list.  The number of elements
	in each list may be different, but (obviously) the amount of data
	stored has to be identical.

	The new passmemdone() function copies data out for the
	CAM_DATA_VADDR and CAM_DATA_SG cases.

	The new passiocleanup() function restores data pointers in
	user CCBs and frees memory.

	Add new functions to support kqueue(2)/kevent(2):

	passreadfilt() tells kevent whether or not the done
	queue is empty.

	passkqfilter() adds a knote to our list.

	passreadfiltdetach() removes a knote from our list.

	Add a new function, passpoll(), for poll(2)/select(2)
	to use.

	Add devstat(9) support for the queued CCB path.

sys/cam/ata/ata_da.c:
	Add support for the BIO_VLIST bio type.

sys/cam/cam_ccb.h:
	Add a new enumeration for the xflags field in the CCB header.
	(This doesn't change the CCB header, just adds an enumeration to
	use.)

sys/cam/cam_xpt.c:
	Add a new function, xpt_setup_ccb_flags(), that allows specifying
	CCB flags.

sys/cam/cam_xpt.h:
	Add a prototype for xpt_setup_ccb_flags().

sys/cam/scsi/scsi_da.c:
	Add support for BIO_VLIST.

sys/dev/md/md.c:
	Add BIO_VLIST support to md(4).

sys/geom/geom_disk.c:
	Add BIO_VLIST support to the GEOM disk class.  Re-factor the I/O size
	limiting code in g_disk_start() a bit.

sys/kern/subr_bus_dma.c:
	Change _bus_dmamap_load_vlist() to take a starting offset and
	length.

	Add a new function, _bus_dmamap_load_pages(), that will load a list
	of physical pages starting at an offset.

	Update _bus_dmamap_load_bio() to allow loading BIO_VLIST bios.
	Allow unmapped I/O to start at an offset.

sys/kern/subr_uio.c:
	Add two new functions, physcopyin_vlist() and physcopyout_vlist().

sys/pc98/include/bus.h:
	Guard kernel-only parts of the pc98 machine/bus.h header with
	#ifdef _KERNEL.

	This allows userland programs to include <machine/bus.h> to get the
	definition of bus_addr_t and bus_size_t.

sys/sys/bio.h:
	Add a new bio flag, BIO_VLIST.

sys/sys/uio.h:
	Add prototypes for physcopyin_vlist() and physcopyout_vlist().

share/man/man4/pass.4:
	Document the CAMIOQUEUE and CAMIOGET ioctls.

usr.sbin/Makefile:
	Add camdd.

usr.sbin/camdd/Makefile:
	Add a makefile for camdd(8).

usr.sbin/camdd/camdd.8:
	Man page for camdd(8).

usr.sbin/camdd/camdd.c:
	The new camdd(8) utility.

Sponsored by:	Spectra Logic
MFC after:	1 week
2015-12-03 20:54:55 +00:00

999 lines
26 KiB
C
Raw Blame History

/*-
* Copyright (c) 2002 Poul-Henning Kamp
* Copyright (c) 2002 Networks Associates Technology, Inc.
* Copyright (c) 2013 The FreeBSD Foundation
* All rights reserved.
*
* This software was developed for the FreeBSD Project by Poul-Henning Kamp
* and NAI Labs, the Security Research Division of Network Associates, Inc.
* under DARPA/SPAWAR contract N66001-01-C-8035 ("CBOSS"), as part of the
* DARPA CHATS research program.
*
* Portions of this software were developed by Konstantin Belousov
* under sponsorship from the FreeBSD Foundation.
*
* 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.
* 3. The names of the authors may not be used to endorse or promote
* products derived from this software without specific prior written
* permission.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR 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 AUTHOR 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/kernel.h>
#include <sys/malloc.h>
#include <sys/bio.h>
#include <sys/ktr.h>
#include <sys/proc.h>
#include <sys/stack.h>
#include <sys/sysctl.h>
#include <sys/vmem.h>
#include <sys/errno.h>
#include <geom/geom.h>
#include <geom/geom_int.h>
#include <sys/devicestat.h>
#include <vm/uma.h>
#include <vm/vm.h>
#include <vm/vm_param.h>
#include <vm/vm_kern.h>
#include <vm/vm_page.h>
#include <vm/vm_object.h>
#include <vm/vm_extern.h>
#include <vm/vm_map.h>
static int g_io_transient_map_bio(struct bio *bp);
static struct g_bioq g_bio_run_down;
static struct g_bioq g_bio_run_up;
static struct g_bioq g_bio_run_task;
/*
* Pace is a hint that we've had some trouble recently allocating
* bios, so we should back off trying to send I/O down the stack
* a bit to let the problem resolve. When pacing, we also turn
* off direct dispatch to also reduce memory pressure from I/Os
* there, at the expxense of some added latency while the memory
* pressures exist. See g_io_schedule_down() for more details
* and limitations.
*/
static volatile u_int pace;
static uma_zone_t biozone;
/*
* The head of the list of classifiers used in g_io_request.
* Use g_register_classifier() and g_unregister_classifier()
* to add/remove entries to the list.
* Classifiers are invoked in registration order.
*/
static TAILQ_HEAD(g_classifier_tailq, g_classifier_hook)
g_classifier_tailq = TAILQ_HEAD_INITIALIZER(g_classifier_tailq);
#include <machine/atomic.h>
static void
g_bioq_lock(struct g_bioq *bq)
{
mtx_lock(&bq->bio_queue_lock);
}
static void
g_bioq_unlock(struct g_bioq *bq)
{
mtx_unlock(&bq->bio_queue_lock);
}
#if 0
static void
g_bioq_destroy(struct g_bioq *bq)
{
mtx_destroy(&bq->bio_queue_lock);
}
#endif
static void
g_bioq_init(struct g_bioq *bq)
{
TAILQ_INIT(&bq->bio_queue);
mtx_init(&bq->bio_queue_lock, "bio queue", NULL, MTX_DEF);
}
static struct bio *
g_bioq_first(struct g_bioq *bq)
{
struct bio *bp;
bp = TAILQ_FIRST(&bq->bio_queue);
if (bp != NULL) {
KASSERT((bp->bio_flags & BIO_ONQUEUE),
("Bio not on queue bp=%p target %p", bp, bq));
bp->bio_flags &= ~BIO_ONQUEUE;
TAILQ_REMOVE(&bq->bio_queue, bp, bio_queue);
bq->bio_queue_length--;
}
return (bp);
}
struct bio *
g_new_bio(void)
{
struct bio *bp;
bp = uma_zalloc(biozone, M_NOWAIT | M_ZERO);
#ifdef KTR
if ((KTR_COMPILE & KTR_GEOM) && (ktr_mask & KTR_GEOM)) {
struct stack st;
CTR1(KTR_GEOM, "g_new_bio(): %p", bp);
stack_save(&st);
CTRSTACK(KTR_GEOM, &st, 3, 0);
}
#endif
return (bp);
}
struct bio *
g_alloc_bio(void)
{
struct bio *bp;
bp = uma_zalloc(biozone, M_WAITOK | M_ZERO);
#ifdef KTR
if ((KTR_COMPILE & KTR_GEOM) && (ktr_mask & KTR_GEOM)) {
struct stack st;
CTR1(KTR_GEOM, "g_alloc_bio(): %p", bp);
stack_save(&st);
CTRSTACK(KTR_GEOM, &st, 3, 0);
}
#endif
return (bp);
}
void
g_destroy_bio(struct bio *bp)
{
#ifdef KTR
if ((KTR_COMPILE & KTR_GEOM) && (ktr_mask & KTR_GEOM)) {
struct stack st;
CTR1(KTR_GEOM, "g_destroy_bio(): %p", bp);
stack_save(&st);
CTRSTACK(KTR_GEOM, &st, 3, 0);
}
#endif
uma_zfree(biozone, bp);
}
struct bio *
g_clone_bio(struct bio *bp)
{
struct bio *bp2;
bp2 = uma_zalloc(biozone, M_NOWAIT | M_ZERO);
if (bp2 != NULL) {
bp2->bio_parent = bp;
bp2->bio_cmd = bp->bio_cmd;
/*
* BIO_ORDERED flag may be used by disk drivers to enforce
* ordering restrictions, so this flag needs to be cloned.
* BIO_UNMAPPED and BIO_VLIST should be inherited, to properly
* indicate which way the buffer is passed.
* Other bio flags are not suitable for cloning.
*/
bp2->bio_flags = bp->bio_flags &
(BIO_ORDERED | BIO_UNMAPPED | BIO_VLIST);
bp2->bio_length = bp->bio_length;
bp2->bio_offset = bp->bio_offset;
bp2->bio_data = bp->bio_data;
bp2->bio_ma = bp->bio_ma;
bp2->bio_ma_n = bp->bio_ma_n;
bp2->bio_ma_offset = bp->bio_ma_offset;
bp2->bio_attribute = bp->bio_attribute;
/* Inherit classification info from the parent */
bp2->bio_classifier1 = bp->bio_classifier1;
bp2->bio_classifier2 = bp->bio_classifier2;
bp->bio_children++;
}
#ifdef KTR
if ((KTR_COMPILE & KTR_GEOM) && (ktr_mask & KTR_GEOM)) {
struct stack st;
CTR2(KTR_GEOM, "g_clone_bio(%p): %p", bp, bp2);
stack_save(&st);
CTRSTACK(KTR_GEOM, &st, 3, 0);
}
#endif
return(bp2);
}
struct bio *
g_duplicate_bio(struct bio *bp)
{
struct bio *bp2;
bp2 = uma_zalloc(biozone, M_WAITOK | M_ZERO);
bp2->bio_flags = bp->bio_flags & (BIO_UNMAPPED | BIO_VLIST);
bp2->bio_parent = bp;
bp2->bio_cmd = bp->bio_cmd;
bp2->bio_length = bp->bio_length;
bp2->bio_offset = bp->bio_offset;
bp2->bio_data = bp->bio_data;
bp2->bio_ma = bp->bio_ma;
bp2->bio_ma_n = bp->bio_ma_n;
bp2->bio_ma_offset = bp->bio_ma_offset;
bp2->bio_attribute = bp->bio_attribute;
bp->bio_children++;
#ifdef KTR
if ((KTR_COMPILE & KTR_GEOM) && (ktr_mask & KTR_GEOM)) {
struct stack st;
CTR2(KTR_GEOM, "g_duplicate_bio(%p): %p", bp, bp2);
stack_save(&st);
CTRSTACK(KTR_GEOM, &st, 3, 0);
}
#endif
return(bp2);
}
void
g_io_init()
{
g_bioq_init(&g_bio_run_down);
g_bioq_init(&g_bio_run_up);
g_bioq_init(&g_bio_run_task);
biozone = uma_zcreate("g_bio", sizeof (struct bio),
NULL, NULL,
NULL, NULL,
0, 0);
}
int
g_io_getattr(const char *attr, struct g_consumer *cp, int *len, void *ptr)
{
struct bio *bp;
int error;
g_trace(G_T_BIO, "bio_getattr(%s)", attr);
bp = g_alloc_bio();
bp->bio_cmd = BIO_GETATTR;
bp->bio_done = NULL;
bp->bio_attribute = attr;
bp->bio_length = *len;
bp->bio_data = ptr;
g_io_request(bp, cp);
error = biowait(bp, "ggetattr");
*len = bp->bio_completed;
g_destroy_bio(bp);
return (error);
}
int
g_io_flush(struct g_consumer *cp)
{
struct bio *bp;
int error;
g_trace(G_T_BIO, "bio_flush(%s)", cp->provider->name);
bp = g_alloc_bio();
bp->bio_cmd = BIO_FLUSH;
bp->bio_flags |= BIO_ORDERED;
bp->bio_done = NULL;
bp->bio_attribute = NULL;
bp->bio_offset = cp->provider->mediasize;
bp->bio_length = 0;
bp->bio_data = NULL;
g_io_request(bp, cp);
error = biowait(bp, "gflush");
g_destroy_bio(bp);
return (error);
}
static int
g_io_check(struct bio *bp)
{
struct g_consumer *cp;
struct g_provider *pp;
off_t excess;
int error;
cp = bp->bio_from;
pp = bp->bio_to;
/* Fail if access counters dont allow the operation */
switch(bp->bio_cmd) {
case BIO_READ:
case BIO_GETATTR:
if (cp->acr == 0)
return (EPERM);
break;
case BIO_WRITE:
case BIO_DELETE:
case BIO_FLUSH:
if (cp->acw == 0)
return (EPERM);
break;
default:
return (EPERM);
}
/* if provider is marked for error, don't disturb. */
if (pp->error)
return (pp->error);
if (cp->flags & G_CF_ORPHAN)
return (ENXIO);
switch(bp->bio_cmd) {
case BIO_READ:
case BIO_WRITE:
case BIO_DELETE:
/* Zero sectorsize or mediasize is probably a lack of media. */
if (pp->sectorsize == 0 || pp->mediasize == 0)
return (ENXIO);
/* Reject I/O not on sector boundary */
if (bp->bio_offset % pp->sectorsize)
return (EINVAL);
/* Reject I/O not integral sector long */
if (bp->bio_length % pp->sectorsize)
return (EINVAL);
/* Reject requests before or past the end of media. */
if (bp->bio_offset < 0)
return (EIO);
if (bp->bio_offset > pp->mediasize)
return (EIO);
/* Truncate requests to the end of providers media. */
excess = bp->bio_offset + bp->bio_length;
if (excess > bp->bio_to->mediasize) {
KASSERT((bp->bio_flags & BIO_UNMAPPED) == 0 ||
round_page(bp->bio_ma_offset +
bp->bio_length) / PAGE_SIZE == bp->bio_ma_n,
("excess bio %p too short", bp));
excess -= bp->bio_to->mediasize;
bp->bio_length -= excess;
if ((bp->bio_flags & BIO_UNMAPPED) != 0) {
bp->bio_ma_n = round_page(bp->bio_ma_offset +
bp->bio_length) / PAGE_SIZE;
}
if (excess > 0)
CTR3(KTR_GEOM, "g_down truncated bio "
"%p provider %s by %d", bp,
bp->bio_to->name, excess);
}
/* Deliver zero length transfers right here. */
if (bp->bio_length == 0) {
CTR2(KTR_GEOM, "g_down terminated 0-length "
"bp %p provider %s", bp, bp->bio_to->name);
return (0);
}
if ((bp->bio_flags & BIO_UNMAPPED) != 0 &&
(bp->bio_to->flags & G_PF_ACCEPT_UNMAPPED) == 0 &&
(bp->bio_cmd == BIO_READ || bp->bio_cmd == BIO_WRITE)) {
if ((error = g_io_transient_map_bio(bp)) >= 0)
return (error);
}
break;
default:
break;
}
return (EJUSTRETURN);
}
/*
* bio classification support.
*
* g_register_classifier() and g_unregister_classifier()
* are used to add/remove a classifier from the list.
* The list is protected using the g_bio_run_down lock,
* because the classifiers are called in this path.
*
* g_io_request() passes bio's that are not already classified
* (i.e. those with bio_classifier1 == NULL) to g_run_classifiers().
* Classifiers can store their result in the two fields
* bio_classifier1 and bio_classifier2.
* A classifier that updates one of the fields should
* return a non-zero value.
* If no classifier updates the field, g_run_classifiers() sets
* bio_classifier1 = BIO_NOTCLASSIFIED to avoid further calls.
*/
int
g_register_classifier(struct g_classifier_hook *hook)
{
g_bioq_lock(&g_bio_run_down);
TAILQ_INSERT_TAIL(&g_classifier_tailq, hook, link);
g_bioq_unlock(&g_bio_run_down);
return (0);
}
void
g_unregister_classifier(struct g_classifier_hook *hook)
{
struct g_classifier_hook *entry;
g_bioq_lock(&g_bio_run_down);
TAILQ_FOREACH(entry, &g_classifier_tailq, link) {
if (entry == hook) {
TAILQ_REMOVE(&g_classifier_tailq, hook, link);
break;
}
}
g_bioq_unlock(&g_bio_run_down);
}
static void
g_run_classifiers(struct bio *bp)
{
struct g_classifier_hook *hook;
int classified = 0;
TAILQ_FOREACH(hook, &g_classifier_tailq, link)
classified |= hook->func(hook->arg, bp);
if (!classified)
bp->bio_classifier1 = BIO_NOTCLASSIFIED;
}
void
g_io_request(struct bio *bp, struct g_consumer *cp)
{
struct g_provider *pp;
struct mtx *mtxp;
int direct, error, first;
KASSERT(cp != NULL, ("NULL cp in g_io_request"));
KASSERT(bp != NULL, ("NULL bp in g_io_request"));
pp = cp->provider;
KASSERT(pp != NULL, ("consumer not attached in g_io_request"));
#ifdef DIAGNOSTIC
KASSERT(bp->bio_driver1 == NULL,
("bio_driver1 used by the consumer (geom %s)", cp->geom->name));
KASSERT(bp->bio_driver2 == NULL,
("bio_driver2 used by the consumer (geom %s)", cp->geom->name));
KASSERT(bp->bio_pflags == 0,
("bio_pflags used by the consumer (geom %s)", cp->geom->name));
/*
* Remember consumer's private fields, so we can detect if they were
* modified by the provider.
*/
bp->_bio_caller1 = bp->bio_caller1;
bp->_bio_caller2 = bp->bio_caller2;
bp->_bio_cflags = bp->bio_cflags;
#endif
if (bp->bio_cmd & (BIO_READ|BIO_WRITE|BIO_GETATTR)) {
KASSERT(bp->bio_data != NULL,
("NULL bp->data in g_io_request(cmd=%hhu)", bp->bio_cmd));
}
if (bp->bio_cmd & (BIO_DELETE|BIO_FLUSH)) {
KASSERT(bp->bio_data == NULL,
("non-NULL bp->data in g_io_request(cmd=%hhu)",
bp->bio_cmd));
}
if (bp->bio_cmd & (BIO_READ|BIO_WRITE|BIO_DELETE)) {
KASSERT(bp->bio_offset % cp->provider->sectorsize == 0,
("wrong offset %jd for sectorsize %u",
bp->bio_offset, cp->provider->sectorsize));
KASSERT(bp->bio_length % cp->provider->sectorsize == 0,
("wrong length %jd for sectorsize %u",
bp->bio_length, cp->provider->sectorsize));
}
g_trace(G_T_BIO, "bio_request(%p) from %p(%s) to %p(%s) cmd %d",
bp, cp, cp->geom->name, pp, pp->name, bp->bio_cmd);
bp->bio_from = cp;
bp->bio_to = pp;
bp->bio_error = 0;
bp->bio_completed = 0;
KASSERT(!(bp->bio_flags & BIO_ONQUEUE),
("Bio already on queue bp=%p", bp));
if ((g_collectstats & G_STATS_CONSUMERS) != 0 ||
((g_collectstats & G_STATS_PROVIDERS) != 0 && pp->stat != NULL))
binuptime(&bp->bio_t0);
else
getbinuptime(&bp->bio_t0);
#ifdef GET_STACK_USAGE
direct = (cp->flags & G_CF_DIRECT_SEND) != 0 &&
(pp->flags & G_PF_DIRECT_RECEIVE) != 0 &&
!g_is_geom_thread(curthread) &&
((pp->flags & G_PF_ACCEPT_UNMAPPED) != 0 ||
(bp->bio_flags & BIO_UNMAPPED) == 0 || THREAD_CAN_SLEEP()) &&
pace == 0;
if (direct) {
/* Block direct execution if less then half of stack left. */
size_t st, su;
GET_STACK_USAGE(st, su);
if (su * 2 > st)
direct = 0;
}
#else
direct = 0;
#endif
if (!TAILQ_EMPTY(&g_classifier_tailq) && !bp->bio_classifier1) {
g_bioq_lock(&g_bio_run_down);
g_run_classifiers(bp);
g_bioq_unlock(&g_bio_run_down);
}
/*
* The statistics collection is lockless, as such, but we
* can not update one instance of the statistics from more
* than one thread at a time, so grab the lock first.
*/
mtxp = mtx_pool_find(mtxpool_sleep, pp);
mtx_lock(mtxp);
if (g_collectstats & G_STATS_PROVIDERS)
devstat_start_transaction(pp->stat, &bp->bio_t0);
if (g_collectstats & G_STATS_CONSUMERS)
devstat_start_transaction(cp->stat, &bp->bio_t0);
pp->nstart++;
cp->nstart++;
mtx_unlock(mtxp);
if (direct) {
error = g_io_check(bp);
if (error >= 0) {
CTR3(KTR_GEOM, "g_io_request g_io_check on bp %p "
"provider %s returned %d", bp, bp->bio_to->name,
error);
g_io_deliver(bp, error);
return;
}
bp->bio_to->geom->start(bp);
} else {
g_bioq_lock(&g_bio_run_down);
first = TAILQ_EMPTY(&g_bio_run_down.bio_queue);
TAILQ_INSERT_TAIL(&g_bio_run_down.bio_queue, bp, bio_queue);
bp->bio_flags |= BIO_ONQUEUE;
g_bio_run_down.bio_queue_length++;
g_bioq_unlock(&g_bio_run_down);
/* Pass it on down. */
if (first)
wakeup(&g_wait_down);
}
}
void
g_io_deliver(struct bio *bp, int error)
{
struct bintime now;
struct g_consumer *cp;
struct g_provider *pp;
struct mtx *mtxp;
int direct, first;
KASSERT(bp != NULL, ("NULL bp in g_io_deliver"));
pp = bp->bio_to;
KASSERT(pp != NULL, ("NULL bio_to in g_io_deliver"));
cp = bp->bio_from;
if (cp == NULL) {
bp->bio_error = error;
bp->bio_done(bp);
return;
}
KASSERT(cp != NULL, ("NULL bio_from in g_io_deliver"));
KASSERT(cp->geom != NULL, ("NULL bio_from->geom in g_io_deliver"));
#ifdef DIAGNOSTIC
/*
* Some classes - GJournal in particular - can modify bio's
* private fields while the bio is in transit; G_GEOM_VOLATILE_BIO
* flag means it's an expected behaviour for that particular geom.
*/
if ((cp->geom->flags & G_GEOM_VOLATILE_BIO) == 0) {
KASSERT(bp->bio_caller1 == bp->_bio_caller1,
("bio_caller1 used by the provider %s", pp->name));
KASSERT(bp->bio_caller2 == bp->_bio_caller2,
("bio_caller2 used by the provider %s", pp->name));
KASSERT(bp->bio_cflags == bp->_bio_cflags,
("bio_cflags used by the provider %s", pp->name));
}
#endif
KASSERT(bp->bio_completed >= 0, ("bio_completed can't be less than 0"));
KASSERT(bp->bio_completed <= bp->bio_length,
("bio_completed can't be greater than bio_length"));
g_trace(G_T_BIO,
"g_io_deliver(%p) from %p(%s) to %p(%s) cmd %d error %d off %jd len %jd",
bp, cp, cp->geom->name, pp, pp->name, bp->bio_cmd, error,
(intmax_t)bp->bio_offset, (intmax_t)bp->bio_length);
KASSERT(!(bp->bio_flags & BIO_ONQUEUE),
("Bio already on queue bp=%p", bp));
/*
* XXX: next two doesn't belong here
*/
bp->bio_bcount = bp->bio_length;
bp->bio_resid = bp->bio_bcount - bp->bio_completed;
#ifdef GET_STACK_USAGE
direct = (pp->flags & G_PF_DIRECT_SEND) &&
(cp->flags & G_CF_DIRECT_RECEIVE) &&
!g_is_geom_thread(curthread);
if (direct) {
/* Block direct execution if less then half of stack left. */
size_t st, su;
GET_STACK_USAGE(st, su);
if (su * 2 > st)
direct = 0;
}
#else
direct = 0;
#endif
/*
* The statistics collection is lockless, as such, but we
* can not update one instance of the statistics from more
* than one thread at a time, so grab the lock first.
*/
if ((g_collectstats & G_STATS_CONSUMERS) != 0 ||
((g_collectstats & G_STATS_PROVIDERS) != 0 && pp->stat != NULL))
binuptime(&now);
mtxp = mtx_pool_find(mtxpool_sleep, cp);
mtx_lock(mtxp);
if (g_collectstats & G_STATS_PROVIDERS)
devstat_end_transaction_bio_bt(pp->stat, bp, &now);
if (g_collectstats & G_STATS_CONSUMERS)
devstat_end_transaction_bio_bt(cp->stat, bp, &now);
cp->nend++;
pp->nend++;
mtx_unlock(mtxp);
if (error != ENOMEM) {
bp->bio_error = error;
if (direct) {
biodone(bp);
} else {
g_bioq_lock(&g_bio_run_up);
first = TAILQ_EMPTY(&g_bio_run_up.bio_queue);
TAILQ_INSERT_TAIL(&g_bio_run_up.bio_queue, bp, bio_queue);
bp->bio_flags |= BIO_ONQUEUE;
g_bio_run_up.bio_queue_length++;
g_bioq_unlock(&g_bio_run_up);
if (first)
wakeup(&g_wait_up);
}
return;
}
if (bootverbose)
printf("ENOMEM %p on %p(%s)\n", bp, pp, pp->name);
bp->bio_children = 0;
bp->bio_inbed = 0;
bp->bio_driver1 = NULL;
bp->bio_driver2 = NULL;
bp->bio_pflags = 0;
g_io_request(bp, cp);
pace = 1;
return;
}
SYSCTL_DECL(_kern_geom);
static long transient_maps;
SYSCTL_LONG(_kern_geom, OID_AUTO, transient_maps, CTLFLAG_RD,
&transient_maps, 0,
"Total count of the transient mapping requests");
u_int transient_map_retries = 10;
SYSCTL_UINT(_kern_geom, OID_AUTO, transient_map_retries, CTLFLAG_RW,
&transient_map_retries, 0,
"Max count of retries used before giving up on creating transient map");
int transient_map_hard_failures;
SYSCTL_INT(_kern_geom, OID_AUTO, transient_map_hard_failures, CTLFLAG_RD,
&transient_map_hard_failures, 0,
"Failures to establish the transient mapping due to retry attempts "
"exhausted");
int transient_map_soft_failures;
SYSCTL_INT(_kern_geom, OID_AUTO, transient_map_soft_failures, CTLFLAG_RD,
&transient_map_soft_failures, 0,
"Count of retried failures to establish the transient mapping");
int inflight_transient_maps;
SYSCTL_INT(_kern_geom, OID_AUTO, inflight_transient_maps, CTLFLAG_RD,
&inflight_transient_maps, 0,
"Current count of the active transient maps");
static int
g_io_transient_map_bio(struct bio *bp)
{
vm_offset_t addr;
long size;
u_int retried;
KASSERT(unmapped_buf_allowed, ("unmapped disabled"));
size = round_page(bp->bio_ma_offset + bp->bio_length);
KASSERT(size / PAGE_SIZE == bp->bio_ma_n, ("Bio too short %p", bp));
addr = 0;
retried = 0;
atomic_add_long(&transient_maps, 1);
retry:
if (vmem_alloc(transient_arena, size, M_BESTFIT | M_NOWAIT, &addr)) {
if (transient_map_retries != 0 &&
retried >= transient_map_retries) {
CTR2(KTR_GEOM, "g_down cannot map bp %p provider %s",
bp, bp->bio_to->name);
atomic_add_int(&transient_map_hard_failures, 1);
return (EDEADLK/* XXXKIB */);
} else {
/*
* Naive attempt to quisce the I/O to get more
* in-flight requests completed and defragment
* the transient_arena.
*/
CTR3(KTR_GEOM, "g_down retrymap bp %p provider %s r %d",
bp, bp->bio_to->name, retried);
pause("g_d_tra", hz / 10);
retried++;
atomic_add_int(&transient_map_soft_failures, 1);
goto retry;
}
}
atomic_add_int(&inflight_transient_maps, 1);
pmap_qenter((vm_offset_t)addr, bp->bio_ma, OFF_TO_IDX(size));
bp->bio_data = (caddr_t)addr + bp->bio_ma_offset;
bp->bio_flags |= BIO_TRANSIENT_MAPPING;
bp->bio_flags &= ~BIO_UNMAPPED;
return (EJUSTRETURN);
}
void
g_io_schedule_down(struct thread *tp __unused)
{
struct bio *bp;
int error;
for(;;) {
g_bioq_lock(&g_bio_run_down);
bp = g_bioq_first(&g_bio_run_down);
if (bp == NULL) {
CTR0(KTR_GEOM, "g_down going to sleep");
msleep(&g_wait_down, &g_bio_run_down.bio_queue_lock,
PRIBIO | PDROP, "-", 0);
continue;
}
CTR0(KTR_GEOM, "g_down has work to do");
g_bioq_unlock(&g_bio_run_down);
if (pace != 0) {
/*
* There has been at least one memory allocation
* failure since the last I/O completed. Pause 1ms to
* give the system a chance to free up memory. We only
* do this once because a large number of allocations
* can fail in the direct dispatch case and there's no
* relationship between the number of these failures and
* the length of the outage. If there's still an outage,
* we'll pause again and again until it's
* resolved. Older versions paused longer and once per
* allocation failure. This was OK for a single threaded
* g_down, but with direct dispatch would lead to max of
* 10 IOPs for minutes at a time when transient memory
* issues prevented allocation for a batch of requests
* from the upper layers.
*
* XXX This pacing is really lame. It needs to be solved
* by other methods. This is OK only because the worst
* case scenario is so rare. In the worst case scenario
* all memory is tied up waiting for I/O to complete
* which can never happen since we can't allocate bios
* for that I/O.
*/
CTR0(KTR_GEOM, "g_down pacing self");
pause("g_down", min(hz/1000, 1));
pace = 0;
}
CTR2(KTR_GEOM, "g_down processing bp %p provider %s", bp,
bp->bio_to->name);
error = g_io_check(bp);
if (error >= 0) {
CTR3(KTR_GEOM, "g_down g_io_check on bp %p provider "
"%s returned %d", bp, bp->bio_to->name, error);
g_io_deliver(bp, error);
continue;
}
THREAD_NO_SLEEPING();
CTR4(KTR_GEOM, "g_down starting bp %p provider %s off %ld "
"len %ld", bp, bp->bio_to->name, bp->bio_offset,
bp->bio_length);
bp->bio_to->geom->start(bp);
THREAD_SLEEPING_OK();
}
}
void
bio_taskqueue(struct bio *bp, bio_task_t *func, void *arg)
{
bp->bio_task = func;
bp->bio_task_arg = arg;
/*
* The taskqueue is actually just a second queue off the "up"
* queue, so we use the same lock.
*/
g_bioq_lock(&g_bio_run_up);
KASSERT(!(bp->bio_flags & BIO_ONQUEUE),
("Bio already on queue bp=%p target taskq", bp));
bp->bio_flags |= BIO_ONQUEUE;
TAILQ_INSERT_TAIL(&g_bio_run_task.bio_queue, bp, bio_queue);
g_bio_run_task.bio_queue_length++;
wakeup(&g_wait_up);
g_bioq_unlock(&g_bio_run_up);
}
void
g_io_schedule_up(struct thread *tp __unused)
{
struct bio *bp;
for(;;) {
g_bioq_lock(&g_bio_run_up);
bp = g_bioq_first(&g_bio_run_task);
if (bp != NULL) {
g_bioq_unlock(&g_bio_run_up);
THREAD_NO_SLEEPING();
CTR1(KTR_GEOM, "g_up processing task bp %p", bp);
bp->bio_task(bp->bio_task_arg);
THREAD_SLEEPING_OK();
continue;
}
bp = g_bioq_first(&g_bio_run_up);
if (bp != NULL) {
g_bioq_unlock(&g_bio_run_up);
THREAD_NO_SLEEPING();
CTR4(KTR_GEOM, "g_up biodone bp %p provider %s off "
"%jd len %ld", bp, bp->bio_to->name,
bp->bio_offset, bp->bio_length);
biodone(bp);
THREAD_SLEEPING_OK();
continue;
}
CTR0(KTR_GEOM, "g_up going to sleep");
msleep(&g_wait_up, &g_bio_run_up.bio_queue_lock,
PRIBIO | PDROP, "-", 0);
}
}
void *
g_read_data(struct g_consumer *cp, off_t offset, off_t length, int *error)
{
struct bio *bp;
void *ptr;
int errorc;
KASSERT(length > 0 && length >= cp->provider->sectorsize &&
length <= MAXPHYS, ("g_read_data(): invalid length %jd",
(intmax_t)length));
bp = g_alloc_bio();
bp->bio_cmd = BIO_READ;
bp->bio_done = NULL;
bp->bio_offset = offset;
bp->bio_length = length;
ptr = g_malloc(length, M_WAITOK);
bp->bio_data = ptr;
g_io_request(bp, cp);
errorc = biowait(bp, "gread");
if (error != NULL)
*error = errorc;
g_destroy_bio(bp);
if (errorc) {
g_free(ptr);
ptr = NULL;
}
return (ptr);
}
int
g_write_data(struct g_consumer *cp, off_t offset, void *ptr, off_t length)
{
struct bio *bp;
int error;
KASSERT(length > 0 && length >= cp->provider->sectorsize &&
length <= MAXPHYS, ("g_write_data(): invalid length %jd",
(intmax_t)length));
bp = g_alloc_bio();
bp->bio_cmd = BIO_WRITE;
bp->bio_done = NULL;
bp->bio_offset = offset;
bp->bio_length = length;
bp->bio_data = ptr;
g_io_request(bp, cp);
error = biowait(bp, "gwrite");
g_destroy_bio(bp);
return (error);
}
int
g_delete_data(struct g_consumer *cp, off_t offset, off_t length)
{
struct bio *bp;
int error;
KASSERT(length > 0 && length >= cp->provider->sectorsize,
("g_delete_data(): invalid length %jd", (intmax_t)length));
bp = g_alloc_bio();
bp->bio_cmd = BIO_DELETE;
bp->bio_done = NULL;
bp->bio_offset = offset;
bp->bio_length = length;
bp->bio_data = NULL;
g_io_request(bp, cp);
error = biowait(bp, "gdelete");
g_destroy_bio(bp);
return (error);
}
void
g_print_bio(struct bio *bp)
{
const char *pname, *cmd = NULL;
if (bp->bio_to != NULL)
pname = bp->bio_to->name;
else
pname = "[unknown]";
switch (bp->bio_cmd) {
case BIO_GETATTR:
cmd = "GETATTR";
printf("%s[%s(attr=%s)]", pname, cmd, bp->bio_attribute);
return;
case BIO_FLUSH:
cmd = "FLUSH";
printf("%s[%s]", pname, cmd);
return;
case BIO_READ:
cmd = "READ";
break;
case BIO_WRITE:
cmd = "WRITE";
break;
case BIO_DELETE:
cmd = "DELETE";
break;
default:
cmd = "UNKNOWN";
printf("%s[%s()]", pname, cmd);
return;
}
printf("%s[%s(offset=%jd, length=%jd)]", pname, cmd,
(intmax_t)bp->bio_offset, (intmax_t)bp->bio_length);
}
Reference in New Issue View Git Blame Copy Permalink