1
0
mirror of https://git.FreeBSD.org/src.git synced 2024-12-22 11:17:19 +00:00
freebsd/sys/geom/concat/g_concat.h
Alexander Motin 40ea77a036 Merge GEOM direct dispatch changes from the projects/camlock branch.
When safety requirements are met, it allows to avoid passing I/O requests
to GEOM g_up/g_down thread, executing them directly in the caller context.
That allows to avoid CPU bottlenecks in g_up/g_down threads, plus avoid
several context switches per I/O.

The defined now safety requirements are:
 - caller should not hold any locks and should be reenterable;
 - callee should not depend on GEOM dual-threaded concurency semantics;
 - on the way down, if request is unmapped while callee doesn't support it,
   the context should be sleepable;
 - kernel thread stack usage should be below 50%.

To keep compatibility with GEOM classes not meeting above requirements
new provider and consumer flags added:
 - G_CF_DIRECT_SEND -- consumer code meets caller requirements (request);
 - G_CF_DIRECT_RECEIVE -- consumer code meets callee requirements (done);
 - G_PF_DIRECT_SEND -- provider code meets caller requirements (done);
 - G_PF_DIRECT_RECEIVE -- provider code meets callee requirements (request).
Capable GEOM class can set them, allowing direct dispatch in cases where
it is safe.  If any of requirements are not met, request is queued to
g_up or g_down thread same as before.

Such GEOM classes were reviewed and updated to support direct dispatch:
CONCAT, DEV, DISK, GATE, MD, MIRROR, MULTIPATH, NOP, PART, RAID, STRIPE,
VFS, ZERO, ZFS::VDEV, ZFS::ZVOL, all classes based on g_slice KPI (LABEL,
MAP, FLASHMAP, etc).

To declare direct completion capability disk(9) KPI got new flag equivalent
to G_PF_DIRECT_SEND -- DISKFLAG_DIRECT_COMPLETION.  da(4) and ada(4) disk
drivers got it set now thanks to earlier CAM locking work.

This change more then twice increases peak block storage performance on
systems with manu CPUs, together with earlier CAM locking changes reaching
more then 1 million IOPS (512 byte raw reads from 16 SATA SSDs on 4 HBAs to
256 user-level threads).

Sponsored by:	iXsystems, Inc.
MFC after:	2 months
2013-10-22 08:22:19 +00:00

128 lines
4.0 KiB
C

/*-
* Copyright (c) 2004-2005 Pawel Jakub Dawidek <pjd@FreeBSD.org>
* 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.
*
* 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.
*
* $FreeBSD$
*/
#ifndef _G_CONCAT_H_
#define _G_CONCAT_H_
#include <sys/endian.h>
#define G_CONCAT_CLASS_NAME "CONCAT"
#define G_CONCAT_MAGIC "GEOM::CONCAT"
/*
* Version history:
* 1 - Initial version number.
* 2 - Added 'stop' command to gconcat(8).
* 3 - Added md_provider field to metadata and '-h' option to gconcat(8).
* 4 - Added md_provsize field to metadata.
*/
#define G_CONCAT_VERSION 4
#ifdef _KERNEL
#define G_CONCAT_TYPE_MANUAL 0
#define G_CONCAT_TYPE_AUTOMATIC 1
#define G_CONCAT_DEBUG(lvl, ...) do { \
if (g_concat_debug >= (lvl)) { \
printf("GEOM_CONCAT"); \
if (g_concat_debug > 0) \
printf("[%u]", lvl); \
printf(": "); \
printf(__VA_ARGS__); \
printf("\n"); \
} \
} while (0)
#define G_CONCAT_LOGREQ(bp, ...) do { \
if (g_concat_debug >= 2) { \
printf("GEOM_CONCAT[2]: "); \
printf(__VA_ARGS__); \
printf(" "); \
g_print_bio(bp); \
printf("\n"); \
} \
} while (0)
struct g_concat_disk {
struct g_consumer *d_consumer;
struct g_concat_softc *d_softc;
off_t d_start;
off_t d_end;
int d_removed;
};
struct g_concat_softc {
u_int sc_type; /* provider type */
struct g_geom *sc_geom;
struct g_provider *sc_provider;
uint32_t sc_id; /* concat unique ID */
struct g_concat_disk *sc_disks;
uint16_t sc_ndisks;
struct mtx sc_lock;
};
#define sc_name sc_geom->name
#endif /* _KERNEL */
struct g_concat_metadata {
char md_magic[16]; /* Magic value. */
uint32_t md_version; /* Version number. */
char md_name[16]; /* Concat name. */
uint32_t md_id; /* Unique ID. */
uint16_t md_no; /* Disk number. */
uint16_t md_all; /* Number of all disks. */
char md_provider[16]; /* Hardcoded provider. */
uint64_t md_provsize; /* Provider's size. */
};
static __inline void
concat_metadata_encode(const struct g_concat_metadata *md, u_char *data)
{
bcopy(md->md_magic, data, sizeof(md->md_magic));
le32enc(data + 16, md->md_version);
bcopy(md->md_name, data + 20, sizeof(md->md_name));
le32enc(data + 36, md->md_id);
le16enc(data + 40, md->md_no);
le16enc(data + 42, md->md_all);
bcopy(md->md_provider, data + 44, sizeof(md->md_provider));
le64enc(data + 60, md->md_provsize);
}
static __inline void
concat_metadata_decode(const u_char *data, struct g_concat_metadata *md)
{
bcopy(data, md->md_magic, sizeof(md->md_magic));
md->md_version = le32dec(data + 16);
bcopy(data + 20, md->md_name, sizeof(md->md_name));
md->md_id = le32dec(data + 36);
md->md_no = le16dec(data + 40);
md->md_all = le16dec(data + 42);
bcopy(data + 44, md->md_provider, sizeof(md->md_provider));
md->md_provsize = le64dec(data + 60);
}
#endif /* _G_CONCAT_H_ */