d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
74f043c943
freebsd-skq/sys/dev/ata/ata-raid.c

1186 lines
34 KiB
C
Raw Normal View History

Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
/*-
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
* Copyright (c) 2000,2001,2002 S<EFBFBD>ren Schmidt <sos@FreeBSD.org>
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
* 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,
* without modification, immediately at the beginning of the file.
* 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 name of the author 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 ``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 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.
*
* $FreeBSD$
*/
#include "opt_global.h"
#include "opt_ata.h"
#include <sys/param.h>
#include <sys/systm.h>
Provide the interface to atacontrol and associated logic. see atacontrol(8) for more. Also the ATA_ENABLE_ATAPI_DMA, ATA_ENABLE_WC and ATA_ENABLE_TAGS options are gone, use the tuneables listed in ata.4 instead from the loader (this makes it possible to switch off DMA before the driver has to touch the devices on broken hardware).
2001-03-15 15:36:25 +00:00
#include <sys/ata.h>
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
#include <sys/kernel.h>
#include <sys/malloc.h>
#include <sys/bio.h>
#include <sys/bus.h>
#include <sys/conf.h>
#include <sys/disk.h>
#include <sys/devicestat.h>
#include <sys/cons.h>
#include <machine/bus.h>
Introduce busspace instead of the good old in/out instructions. Not pretty but it works (I hope)...
2001-02-06 16:44:25 +00:00
#include <sys/rman.h>
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
#include <dev/ata/ata-all.h>
#include <dev/ata/ata-disk.h>
#include <dev/ata/ata-raid.h>
/* device structures */
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
static d_open_t aropen;
static d_strategy_t arstrategy;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
static struct cdevsw ar_cdevsw = {
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
/* open */ aropen,
/* close */ nullclose,
/* read */ physread,
/* write */ physwrite,
/* ioctl */ noioctl,
/* poll */ nopoll,
/* mmap */ nommap,
/* strategy */ arstrategy,
/* name */ "ar",
/* maj */ 157,
/* dump */ nodump,
/* psize */ nopsize,
/* flags */ D_DISK,
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
};
static struct cdevsw ardisk_cdevsw;
/* prototypes */
static void ar_done(struct bio *);
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
static void ar_config_changed(struct ar_softc *, int);
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
static int ar_rebuild(struct ar_softc *);
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
static int ar_highpoint_read_conf(struct ad_softc *, struct ar_softc **);
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
static int ar_highpoint_write_conf(struct ar_softc *);
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
static int ar_promise_read_conf(struct ad_softc *, struct ar_softc **);
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
static int ar_promise_write_conf(struct ar_softc *);
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
static int ar_rw(struct ad_softc *, u_int32_t, int, caddr_t, int);
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
/* misc defines */
#define AD_STRATEGY(x) si_disk->d_devsw->d_strategy(x)
#define AD_SOFTC(x) ((struct ad_softc *)(x.device->driver))
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
#define AR_READ 0x01
#define AR_WRITE 0x02
#define AR_WAIT 0x04
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
/* internal vars */
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
static struct ar_softc **ar_table = NULL;
Staticize some malloc M_ instances.
2000-12-08 20:09:00 +00:00
static MALLOC_DEFINE(M_AR, "AR driver", "ATA RAID driver");
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
int
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
ata_raiddisk_attach(struct ad_softc *adp)
{
struct ar_softc *rdp;
int array, disk;
switch(adp->device->channel->chiptype) {
default:
return 0;
case 0x4d33105a: case 0x4d38105a: case 0x4d30105a:
case 0x0d30105a: case 0x4d68105a: case 0x6268105a:
case 0x00041103: case 0x00051103: case 0x00081103:
}
if (ar_table) {
for (array = 0; array < MAX_ARRAYS; array++) {
if (!(rdp = ar_table[array]) || !rdp->flags)
continue;
for (disk = 0; disk < rdp->total_disks; disk++) {
if (rdp->disks[disk].device == adp->device) {
ata_prtdev(rdp->disks[disk].device,
"inserted into ar%d disk%d as spare\n",
array, disk);
rdp->disks[disk].flags = (AR_DF_PRESENT | AR_DF_SPARE);
ar_config_changed(rdp, 1);
return 1;
}
}
}
}
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
if (!ar_table)
ar_table = malloc(sizeof(struct ar_soft *) * MAX_ARRAYS,
M_AR, M_NOWAIT | M_ZERO);
if (!ar_table) {
ata_prtdev(adp->device, "no memory for ATA raid array\n");
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
return 0;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
switch(adp->device->channel->chiptype) {
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
case 0x4d33105a: case 0x4d38105a: case 0x4d30105a:
case 0x0d30105a: case 0x4d68105a: case 0x6268105a:
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
/* test RAID bit in PCI reg XXX */
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
return (ar_promise_read_conf(adp, ar_table));
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
case 0x00041103: case 0x00051103: case 0x00081103:
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
return (ar_highpoint_read_conf(adp, ar_table));
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
return 0;
}
int
ata_raiddisk_detach(struct ad_softc *adp)
{
struct ar_softc *rdp;
int array, disk;
switch(adp->device->channel->chiptype) {
default:
return 0;
case 0x4d33105a: case 0x4d38105a: case 0x4d30105a:
case 0x0d30105a: case 0x4d68105a: case 0x6268105a:
case 0x00041103: case 0x00051103: case 0x00081103:
}
if (ar_table) {
for (array = 0; array < MAX_ARRAYS; array++) {
if (!(rdp = ar_table[array]) || !rdp->flags)
continue;
for (disk = 0; disk < rdp->total_disks; disk++) {
if (rdp->disks[disk].device == adp->device) {
ata_prtdev(rdp->disks[disk].device,
"deleted from ar%d disk%d\n", array, disk);
rdp->disks[disk].flags &= ~(AR_DF_PRESENT | AR_DF_ONLINE);
ar_config_changed(rdp, 1);
return 1;
}
}
}
}
return 0;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
void
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
ata_raid_attach()
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
{
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
struct ar_softc *rdp;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
dev_t dev;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
int array, disk;
Dont try to attach ATA RAID's if none found.
2002-02-16 08:10:24 +00:00
if (!ar_table)
return;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
for (array = 0; array < MAX_ARRAYS; array++) {
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
if (!(rdp = ar_table[array]) || !rdp->flags)
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
continue;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
ar_config_changed(rdp, 0);
dev = disk_create(rdp->lun, &rdp->disk, 0, &ar_cdevsw,&ardisk_cdevsw);
dev->si_drv1 = rdp;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
dev->si_iosize_max = 256 * DEV_BSIZE;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
rdp->dev = dev;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
printf("ar%d: %lluMB <ATA ",
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
rdp->lun, (unsigned long long)
Fixed a printf format error again. Rev.127 was clobbered in rev.1.128 by removing parentheses. The main bug is in gcc: on machines with 64-bit longs and 64-bit long longs, (unsigned long long)rdp->total_sectors / ((1024L * 1024L) / DEV_BSIZE)) has type plain unsigned long instead of the correctly promoted type unsigned long long, so it can not be printfed using %llu format. Even 1ULL / 1L is mispromoted. Anyway, casting the correct operand automatically avoids the problem. We do not want to to pessimize the division; we just want to convert to a common maximal type for printing.
2002-03-06 06:33:33 +00:00
(rdp->total_sectors / ((1024L * 1024L) / DEV_BSIZE)));
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
switch (rdp->flags & (AR_F_RAID0 | AR_F_RAID1 | AR_F_SPAN)) {
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
case AR_F_RAID0:
printf("RAID0 "); break;
case AR_F_RAID1:
printf("RAID1 "); break;
case AR_F_SPAN:
printf("SPAN "); break;
case (AR_F_RAID0 | AR_F_RAID1):
printf("RAID0+1 "); break;
default:
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
printf("unknown 0x%x> ", rdp->flags);
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
return;
}
printf("array> [%d/%d/%d] status: ",
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
rdp->cylinders, rdp->heads, rdp->sectors);
switch (rdp->flags & (AR_F_DEGRADED | AR_F_READY)) {
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
case AR_F_READY:
printf("READY");
break;
case (AR_F_DEGRADED | AR_F_READY):
printf("DEGRADED");
break;
default:
printf("BROKEN");
break;
}
printf(" subdisks:\n");
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
for (disk = 0; disk < rdp->total_disks; disk++) {
if (rdp->disks[disk].flags & AR_DF_ONLINE)
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
printf(" %d READY ", disk);
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
else if (rdp->disks[disk].flags & AR_DF_ASSIGNED)
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
printf(" %d DOWN ", disk);
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
else if (rdp->disks[disk].flags & AR_DF_SPARE)
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
printf(" %d SPARE ", disk);
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
else if (rdp->disks[disk].flags & AR_DF_PRESENT)
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
printf(" %d FREE ", disk);
else
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
printf(" %d INVALID no RAID config info on this disk\n", disk);
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
if (rdp->disks[disk].flags & AR_DF_PRESENT)
ad_print(AD_SOFTC(rdp->disks[disk]));
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
}
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
}
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
int
ata_raid_rebuild(int array)
{
struct ar_softc *rdp;
if (!ar_table || !(rdp = ar_table[array]))
return ENXIO;
return ar_rebuild(rdp);
}
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
static int
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
aropen(dev_t dev, int flags, int fmt, struct thread *td)
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
{
struct ar_softc *rdp = dev->si_drv1;
struct disklabel *dl;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
dl = &rdp->disk.d_label;
bzero(dl, sizeof *dl);
dl->d_secsize = DEV_BSIZE;
dl->d_nsectors = rdp->sectors;
dl->d_ntracks = rdp->heads;
dl->d_ncylinders = rdp->cylinders;
dl->d_secpercyl = rdp->sectors * rdp->heads;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
dl->d_secperunit = rdp->total_sectors;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
return 0;
}
static void
arstrategy(struct bio *bp)
{
struct ar_softc *rdp = bp->bio_dev->si_drv1;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
int blkno, count, chunk, lba, lbs, tmplba;
int drv = 0, change = 0;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
caddr_t data;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
if (!(rdp->flags & AR_F_READY)) {
bp->bio_flags |= BIO_ERROR;
bp->bio_error = EIO;
biodone(bp);
return;
}
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
bp->bio_resid = bp->bio_bcount;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
blkno = bp->bio_pblkno;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
data = bp->bio_data;
for (count = howmany(bp->bio_bcount, DEV_BSIZE); count > 0;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
count -= chunk, blkno += chunk, data += (chunk * DEV_BSIZE)) {
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
struct ar_buf *buf1, *buf2;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
switch (rdp->flags & (AR_F_RAID0 | AR_F_RAID1 | AR_F_SPAN)) {
case AR_F_SPAN:
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
lba = blkno;
while (lba >= AD_SOFTC(rdp->disks[drv])->total_secs-rdp->reserved)
lba -= AD_SOFTC(rdp->disks[drv++])->total_secs-rdp->reserved;
chunk = min(AD_SOFTC(rdp->disks[drv])->total_secs-rdp->reserved-lba,
count);
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
break;
case AR_F_RAID0:
case AR_F_RAID0 | AR_F_RAID1:
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
tmplba = blkno / rdp->interleave;
chunk = blkno % rdp->interleave;
if (tmplba == rdp->total_sectors / rdp->interleave) {
lbs = (rdp->total_sectors-(tmplba*rdp->interleave))/rdp->width;
drv = chunk / lbs;
lba = ((tmplba/rdp->width)*rdp->interleave) + chunk%lbs;
chunk = min(count, lbs);
Handle the case where the last piece of a RAID0 (striped) disk is not of interleave size.
2001-03-21 11:48:14 +00:00
}
else {
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
drv = tmplba % rdp->width;
lba = ((tmplba / rdp->width) * rdp->interleave) + chunk;
Handle the case where the last piece of a RAID0 (striped) disk is not of interleave size.
2001-03-21 11:48:14 +00:00
chunk = min(count, rdp->interleave - chunk);
}
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
break;
case AR_F_RAID1:
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
drv = 0;
lba = blkno;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
chunk = count;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
break;
default:
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
printf("ar%d: unknown array type in arstrategy\n", rdp->lun);
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
bp->bio_flags |= BIO_ERROR;
bp->bio_error = EIO;
biodone(bp);
return;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
buf1 = malloc(sizeof(struct ar_buf), M_AR, M_NOWAIT | M_ZERO);
buf1->bp.bio_pblkno = lba;
if ((buf1->drive = drv) > 0)
Only offset raid disks > 1 on the HPT, this should solve the boot problems.. However this demands that dangerously dedicated disks use an offset of at least 10 from the start to not overwrite the raid config sector on the HPT...
2000-11-12 20:45:14 +00:00
buf1->bp.bio_pblkno += rdp->offset;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
buf1->bp.bio_caller1 = (void *)rdp;
buf1->bp.bio_bcount = chunk * DEV_BSIZE;
buf1->bp.bio_data = data;
buf1->bp.bio_cmd = bp->bio_cmd;
buf1->bp.bio_flags = bp->bio_flags;
buf1->bp.bio_done = ar_done;
buf1->org = bp;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
switch (rdp->flags & (AR_F_RAID0 | AR_F_RAID1 | AR_F_SPAN)) {
case AR_F_SPAN:
case AR_F_RAID0:
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
if (rdp->disks[buf1->drive].flags & AR_DF_ONLINE &&
!AD_SOFTC(rdp->disks[buf1->drive])->dev->si_disk) {
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
rdp->disks[buf1->drive].flags &= ~AR_DF_ONLINE;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
ar_config_changed(rdp, 1);
free(buf1, M_AR);
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
bp->bio_flags |= BIO_ERROR;
bp->bio_error = EIO;
biodone(bp);
return;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
buf1->bp.bio_dev = AD_SOFTC(rdp->disks[buf1->drive])->dev;
buf1->bp.bio_dev->AD_STRATEGY((struct bio *)buf1);
break;
case AR_F_RAID1:
case AR_F_RAID0 | AR_F_RAID1:
if (rdp->flags & AR_F_REBUILDING) {
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
int start = rdp->lock_start / rdp->width;
int end = rdp->lock_end / rdp->width;
if ((bp->bio_pblkno >= end && bp->bio_pblkno < end) ||
((bp->bio_pblkno + chunk) >= start &&
(bp->bio_pblkno + chunk) < end)) {
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
tsleep(rdp, PRIBIO, "arwait", 0);
}
}
if (rdp->disks[buf1->drive].flags & AR_DF_ONLINE &&
!AD_SOFTC(rdp->disks[buf1->drive])->dev->si_disk) {
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
rdp->disks[buf1->drive].flags &= ~AR_DF_ONLINE;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
change = 1;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
}
if (rdp->disks[buf1->drive + rdp->width].flags & AR_DF_ONLINE &&
!AD_SOFTC(rdp->disks[buf1->drive + rdp->width])->dev->si_disk) {
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
rdp->disks[buf1->drive + rdp->width].flags &= ~AR_DF_ONLINE;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
change = 1;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
}
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
if (change)
ar_config_changed(rdp, 1);
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
if (!(rdp->flags & AR_F_READY)) {
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
free(buf1, M_AR);
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
bp->bio_flags |= BIO_ERROR;
bp->bio_error = EIO;
biodone(bp);
return;
}
Misc little cleanups: Link if only ATAPI device in kernel config Remove unused #includes Rearrange a bit in ata-raid to make diff against -stable smaller Enable wc as default again, dunne how this happend...
2002-03-05 09:24:19 +00:00
if (bp->bio_cmd == BIO_READ) {
if ((buf1->bp.bio_pblkno <
(rdp->disks[buf1->drive].last_lba - AR_PROXIMITY) ||
buf1->bp.bio_pblkno >
(rdp->disks[buf1->drive].last_lba + AR_PROXIMITY) ||
!(rdp->disks[buf1->drive].flags & AR_DF_ONLINE)) &&
(rdp->disks[buf1->drive+rdp->width].flags & AR_DF_ONLINE))
buf1->drive = buf1->drive + rdp->width;
}
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
if (bp->bio_cmd == BIO_WRITE) {
if (rdp->disks[buf1->drive + rdp->width].flags & AR_DF_ONLINE) {
if (rdp->disks[buf1->drive].flags & AR_DF_ONLINE) {
buf2 = malloc(sizeof(struct ar_buf), M_AR, M_NOWAIT);
bcopy(buf1, buf2, sizeof(struct ar_buf));
buf1->mirror = buf2;
buf2->mirror = buf1;
buf2->drive = buf1->drive + rdp->width;
buf2->bp.bio_dev =
AD_SOFTC(rdp->disks[buf2->drive])->dev;
buf2->bp.bio_dev->AD_STRATEGY((struct bio *)buf2);
rdp->disks[buf2->drive].last_lba =
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
buf2->bp.bio_pblkno + chunk;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
}
else
buf1->drive = buf1->drive + rdp->width;
}
}
buf1->bp.bio_dev = AD_SOFTC(rdp->disks[buf1->drive])->dev;
buf1->bp.bio_dev->AD_STRATEGY((struct bio *)buf1);
rdp->disks[buf1->drive].last_lba = buf1->bp.bio_pblkno + chunk;
break;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
default:
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
printf("ar%d: unknown array type in arstrategy\n", rdp->lun);
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
}
}
static void
ar_done(struct bio *bp)
{
struct ar_softc *rdp = (struct ar_softc *)bp->bio_caller1;
struct ar_buf *buf = (struct ar_buf *)bp;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
switch (rdp->flags & (AR_F_RAID0 | AR_F_RAID1 | AR_F_SPAN)) {
case AR_F_SPAN:
case AR_F_RAID0:
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
if (buf->bp.bio_flags & BIO_ERROR) {
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
rdp->disks[buf->drive].flags &= ~AR_DF_ONLINE;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
ar_config_changed(rdp, 1);
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
buf->org->bio_flags |= BIO_ERROR;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
buf->org->bio_error = EIO;
biodone(buf->org);
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
else {
buf->org->bio_resid -= buf->bp.bio_bcount;
if (buf->org->bio_resid == 0)
biodone(buf->org);
}
break;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
case AR_F_RAID1:
case AR_F_RAID0 | AR_F_RAID1:
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
if (buf->bp.bio_flags & BIO_ERROR) {
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
rdp->disks[buf->drive].flags &= ~AR_DF_ONLINE;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
ar_config_changed(rdp, 1);
if (rdp->flags & AR_F_READY) {
if (buf->bp.bio_cmd == BIO_READ) {
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
if (buf->drive < rdp->width)
buf->drive = buf->drive + rdp->width;
else
buf->drive = buf->drive - rdp->width;
buf->bp.bio_dev = AD_SOFTC(rdp->disks[buf->drive])->dev;
buf->bp.bio_flags = buf->org->bio_flags;
buf->bp.bio_error = 0;
buf->bp.bio_dev->AD_STRATEGY((struct bio *)buf);
return;
}
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
if (buf->bp.bio_cmd == BIO_WRITE) {
if (buf->flags & AB_F_DONE) {
buf->org->bio_resid -= buf->bp.bio_bcount;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
if (buf->org->bio_resid == 0)
biodone(buf->org);
}
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
else
buf->mirror->flags |= AB_F_DONE;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
}
}
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
else {
buf->org->bio_flags |= BIO_ERROR;
buf->org->bio_error = EIO;
biodone(buf->org);
}
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
}
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
else {
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
if (buf->bp.bio_cmd == BIO_WRITE) {
if (buf->mirror && !(buf->flags & AB_F_DONE)){
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
buf->mirror->flags |= AB_F_DONE;
break;
}
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
buf->org->bio_resid -= buf->bp.bio_bcount;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
if (buf->org->bio_resid == 0)
biodone(buf->org);
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
break;
default:
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
printf("ar%d: unknown array type in ar_done\n", rdp->lun);
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
free(buf, M_AR);
}
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
static void
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
ar_config_changed(struct ar_softc *rdp, int writeback)
{
int disk, flags;
flags = rdp->flags;
rdp->flags |= AR_F_READY;
rdp->flags &= ~AR_F_DEGRADED;
for (disk = 0; disk < rdp->total_disks; disk++) {
switch (rdp->flags & (AR_F_RAID0 | AR_F_RAID1 | AR_F_SPAN)) {
case AR_F_SPAN:
case AR_F_RAID0:
if (!(rdp->disks[disk].flags & AR_DF_ONLINE)) {
rdp->flags &= ~AR_F_READY;
printf("ar%d: ERROR - array broken\n", rdp->lun);
}
break;
case AR_F_RAID1:
case AR_F_RAID0 | AR_F_RAID1:
if (disk < rdp->width) {
if (!(rdp->disks[disk].flags & AR_DF_ONLINE) &&
!(rdp->disks[disk + rdp->width].flags & AR_DF_ONLINE)) {
rdp->flags &= ~AR_F_READY;
printf("ar%d: ERROR - array broken\n", rdp->lun);
}
else if (((rdp->disks[disk].flags & AR_DF_ONLINE) &&
!(rdp->disks
[disk + rdp->width].flags & AR_DF_ONLINE))||
(!(rdp->disks[disk].flags & AR_DF_ONLINE) &&
(rdp->disks
[disk + rdp->width].flags & AR_DF_ONLINE))) {
rdp->flags |= AR_F_DEGRADED;
if (!(flags & AR_F_DEGRADED))
printf("ar%d: WARNING - mirror lost\n", rdp->lun);
}
}
break;
}
if ((rdp->disks[disk].flags & AR_DF_PRESENT) &&
rdp->disks[disk].device) {
if (rdp->disks[disk].flags & AR_DF_ONLINE)
ata_drawerleds(rdp->disks[disk].device, ATA_LED_GREEN);
else
ata_drawerleds(rdp->disks[disk].device, ATA_LED_RED);
}
}
if (writeback) {
if (rdp->flags & AR_F_PROMISE_RAID)
ar_promise_write_conf(rdp);
if (rdp->flags & AR_F_HIGHPOINT_RAID)
ar_highpoint_write_conf(rdp);
}
}
static int
ar_rebuild(struct ar_softc *rdp)
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
{
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
caddr_t buffer;
int count = 0;
int disk;
if ((rdp->flags & (AR_F_READY|AR_F_DEGRADED)) != (AR_F_READY|AR_F_DEGRADED))
return EEXIST;
for (disk = 0; disk < rdp->total_disks; disk++) {
if (((rdp->disks[disk].flags&(AR_DF_PRESENT|AR_DF_ONLINE|AR_DF_SPARE))==
(AR_DF_PRESENT | AR_DF_SPARE)) && rdp->disks[disk].device) {
if (AD_SOFTC(rdp->disks[disk])->total_secs <
rdp->disks[disk].disk_sectors) {
ata_prtdev(rdp->disks[disk].device,
"disk capacity too small for this RAID config\n");
rdp->disks[disk].flags &= ~AR_DF_SPARE;
continue;
}
ata_drawerleds(rdp->disks[disk].device, ATA_LED_ORANGE);
count++;
}
}
if (!count)
return ENODEV;
/* setup start conditions */
rdp->lock_start = 0;
rdp->lock_end = rdp->lock_start + 256;
rdp->flags |= AR_F_REBUILDING;
buffer = malloc(256 * DEV_BSIZE, M_AR, M_NOWAIT | M_ZERO);
/* now go copy entire disk(s) */
Forgot this litte patch, scale the individual disk size from the arrays total size depending on width of the array.
2002-03-03 16:36:19 +00:00
while (rdp->lock_start < (rdp->total_sectors / rdp->width)) {
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
for (disk = 0; disk < rdp->width; disk++) {
if (((rdp->disks[disk].flags & AR_DF_ONLINE) &&
(rdp->disks[disk + rdp->width].flags & AR_DF_ONLINE)) ||
((rdp->disks[disk].flags & AR_DF_ONLINE) &&
!(rdp->disks[disk + rdp->width].flags & AR_DF_SPARE)) ||
((rdp->disks[disk + rdp->width].flags & AR_DF_ONLINE) &&
!(rdp->disks[disk].flags & AR_DF_SPARE)))
continue;
if (rdp->disks[disk].flags & AR_DF_ONLINE)
ar_rw(AD_SOFTC(rdp->disks[disk]), rdp->lock_start,
256 * DEV_BSIZE, buffer, AR_READ | AR_WAIT);
else
ar_rw(AD_SOFTC(rdp->disks[disk + rdp->width]), rdp->lock_start,
256 * DEV_BSIZE, buffer, AR_READ | AR_WAIT);
if (rdp->disks[disk].flags & AR_DF_ONLINE)
ar_rw(AD_SOFTC(rdp->disks[disk + rdp->width]), rdp->lock_start,
256 * DEV_BSIZE, buffer, AR_WRITE | AR_WAIT);
else
ar_rw(AD_SOFTC(rdp->disks[disk]), rdp->lock_start,
256 * DEV_BSIZE, buffer, AR_WRITE | AR_WAIT);
}
rdp->lock_start = rdp->lock_end;
Forgot this litte patch, scale the individual disk size from the arrays total size depending on width of the array.
2002-03-03 16:36:19 +00:00
rdp->lock_end = rdp->lock_start +
min(256, (rdp->total_sectors / rdp->width) - rdp->lock_end);
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
wakeup(rdp);
}
free(buffer, M_AR);
for (disk = 0; disk < rdp->total_disks; disk++) {
if ((rdp->disks[disk].flags&(AR_DF_PRESENT|AR_DF_ONLINE|AR_DF_SPARE))==
(AR_DF_PRESENT | AR_DF_SPARE)) {
rdp->disks[disk].flags &= ~AR_DF_SPARE;
rdp->disks[disk].flags |= (AR_DF_ASSIGNED | AR_DF_ONLINE);
}
}
rdp->flags &= ~AR_F_REBUILDING;
ar_config_changed(rdp, 1);
return 0;
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
}
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
static int
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
ar_highpoint_read_conf(struct ad_softc *adp, struct ar_softc **raidp)
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
{
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
struct highpoint_raid_conf *info;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
struct ar_softc *raid = NULL;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
int array, disk_number = 0, retval = 0;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
if (!(info = (struct highpoint_raid_conf *)
malloc(sizeof(struct highpoint_raid_conf), M_AR, M_NOWAIT | M_ZERO)))
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
return retval;
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
if (ar_rw(adp, HPT_LBA, sizeof(struct highpoint_raid_conf),
(caddr_t)info, AR_READ | AR_WAIT)) {
Minor changes to the ATA RAID support code, remove some verbosity and put some under bootverbose..
2000-10-18 18:49:42 +00:00
if (bootverbose)
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
printf("ar: HighPoint read conf failed\n");
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
goto highpoint_out;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
/* check if this is a HighPoint RAID struct */
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
if (info->magic != HPT_MAGIC_OK && info->magic != HPT_MAGIC_BAD) {
Minor changes to the ATA RAID support code, remove some verbosity and put some under bootverbose..
2000-10-18 18:49:42 +00:00
if (bootverbose)
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
printf("ar: HighPoint check1 failed\n");
goto highpoint_out;
}
/* is this disk defined, or an old leftover/spare ? */
if (!info->magic_0) {
if (bootverbose)
printf("ar: HighPoint check2 failed\n");
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
goto highpoint_out;
Update the config gathering code for both Promise & HPT
2000-11-01 17:35:44 +00:00
}
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
/* now convert HighPoint config info into our generic form */
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
for (array = 0; array < MAX_ARRAYS; array++) {
Update the config gathering code for both Promise & HPT
2000-11-01 17:35:44 +00:00
if (!raidp[array]) {
raidp[array] =
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
(struct ar_softc*)malloc(sizeof(struct ar_softc), M_AR,
Use M_ZERO flag on malloc when approbiate.
2000-12-26 12:05:49 +00:00
M_NOWAIT | M_ZERO);
Update the config gathering code for both Promise & HPT
2000-11-01 17:35:44 +00:00
if (!raidp[array]) {
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
printf("ar%d: failed to allocate raid config storage\n", array);
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
goto highpoint_out;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
}
Update the config gathering code for both Promise & HPT
2000-11-01 17:35:44 +00:00
raid = raidp[array];
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
if (raid->flags & AR_F_PROMISE_RAID)
continue;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
switch (info->type) {
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
case HPT_T_RAID0:
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
case HPT_T_RAID01_RAID0:
Clean up the raid code a bit, also allow disks on HPT controllers to be swapped around and still be put in the correct order in a raid.
2000-10-22 12:17:57 +00:00
/* check the order byte to determine what this really is */
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
switch (info->order) {
case HPT_O_DOWN:
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
if (raid->magic_0 && raid->magic_0 != info->magic_0)
Clean up the raid code a bit, also allow disks on HPT controllers to be swapped around and still be put in the correct order in a raid.
2000-10-22 12:17:57 +00:00
continue;
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
raid->magic_0 = info->magic_0;
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
raid->magic_1 = info->magic_1;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
raid->flags |= AR_F_RAID0;
raid->interleave = 1 << info->stripe_shift;
disk_number = info->disk_number;
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
info->magic = 0; /* mark bad */
Clean up the raid code a bit, also allow disks on HPT controllers to be swapped around and still be put in the correct order in a raid.
2000-10-22 12:17:57 +00:00
break;
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
case HPT_O_RAID01DEGRADED:
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
if (raid->magic_0 && raid->magic_0 != info->magic_0)
Clean up the raid code a bit, also allow disks on HPT controllers to be swapped around and still be put in the correct order in a raid.
2000-10-22 12:17:57 +00:00
continue;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
raid->magic_0 = info->magic_0;
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
raid->magic_1 = info->magic_1;
raid->flags |= AR_F_RAID0;
raid->interleave = 1 << info->stripe_shift;
disk_number = info->disk_number;
Clean up the raid code a bit, also allow disks on HPT controllers to be swapped around and still be put in the correct order in a raid.
2000-10-22 12:17:57 +00:00
break;
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
case HPT_O_RAID01SRC:
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
if ((raid->magic_0 && raid->magic_0 != info->magic_0) ||
(raid->magic_1 && raid->magic_1 != info->magic_1))
Clean up the raid code a bit, also allow disks on HPT controllers to be swapped around and still be put in the correct order in a raid.
2000-10-22 12:17:57 +00:00
continue;
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
raid->magic_0 = info->magic_0;
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
raid->magic_1 = info->magic_1;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
raid->flags |= (AR_F_RAID0 | AR_F_RAID1);
raid->interleave = 1 << info->stripe_shift;
disk_number = info->disk_number;
break;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
case HPT_O_RAID01DST:
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
if (raid->magic_1 && raid->magic_1 != info->magic_1)
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
continue;
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
raid->magic_1 = info->magic_1;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
raid->flags |= (AR_F_RAID0 | AR_F_RAID1);
raid->interleave = 1 << info->stripe_shift;
disk_number = info->disk_number + info->array_width;
break;
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
case HPT_O_READY:
if (raid->magic_0 && raid->magic_0 != info->magic_0)
continue;
raid->magic_0 = info->magic_0;
raid->flags |= AR_F_RAID0;
raid->interleave = 1 << info->stripe_shift;
disk_number = info->disk_number;
break;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
break;
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
case HPT_T_RAID1:
if (raid->magic_0 && raid->magic_0 != info->magic_0)
continue;
raid->magic_0 = info->magic_0;
raid->flags |= AR_F_RAID1;
disk_number = (info->disk_number > 0);
break;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
case HPT_T_SPAN:
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
if (raid->magic_0 && raid->magic_0 != info->magic_0)
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
continue;
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
raid->magic_0 = info->magic_0;
Clean up the raid code a bit, also allow disks on HPT controllers to be swapped around and still be put in the correct order in a raid.
2000-10-22 12:17:57 +00:00
raid->flags |= AR_F_SPAN;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
disk_number = info->disk_number;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
break;
default:
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
printf("ar%d: HighPoint unknown RAID type 0x%02x\n",
array, info->type);
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
goto highpoint_out;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
raid->flags |= AR_F_HIGHPOINT_RAID;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
raid->disks[disk_number].device = adp->device;
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
raid->disks[disk_number].flags = (AR_DF_PRESENT | AR_DF_ASSIGNED);
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
if (info->magic == HPT_MAGIC_OK) {
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
raid->disks[disk_number].flags |= AR_DF_ONLINE;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
raid->flags |= AR_F_READY;
Update the config gathering code for both Promise & HPT
2000-11-01 17:35:44 +00:00
raid->lun = array;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
raid->width = info->array_width;
Update the config gathering code for both Promise & HPT
2000-11-01 17:35:44 +00:00
raid->heads = 255;
raid->sectors = 63;
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
raid->cylinders = (info->total_sectors - HPT_LBA) / (63 * 255);
raid->total_sectors = info->total_sectors - (HPT_LBA * raid->width);
Update the config gathering code for both Promise & HPT
2000-11-01 17:35:44 +00:00
raid->offset = 10;
raid->reserved = 10;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
raid->disks[disk_number].disk_sectors =
info->total_sectors / info->array_width;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
else
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
raid->disks[disk_number].flags &= ~ AR_DF_ONLINE;
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
if ((raid->flags & AR_F_RAID0) && (raid->total_disks < raid->width))
raid->total_disks = raid->width;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
if (disk_number >= raid->total_disks)
raid->total_disks = disk_number + 1;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
retval = 1;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
break;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
highpoint_out:
free(info, M_AR);
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
return retval;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
static int
ar_highpoint_write_conf(struct ar_softc *rdp)
{
struct highpoint_raid_conf *config;
struct timeval timestamp;
int disk;
microtime(&timestamp);
rdp->magic_0 = timestamp.tv_sec + 1;
rdp->magic_1 = timestamp.tv_sec;
for (disk = 0; disk < rdp->total_disks; disk++) {
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
if (!(config = (struct highpoint_raid_conf *)
malloc(sizeof(struct highpoint_raid_conf),
M_AR, M_NOWAIT | M_ZERO))) {
printf("ar%d: Highpoint write conf failed\n", rdp->lun);
return -1;
}
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
if ((rdp->disks[disk].flags & (AR_DF_PRESENT | AR_DF_ONLINE)) ==
(AR_DF_PRESENT | AR_DF_ONLINE))
config->magic = HPT_MAGIC_OK;
if (rdp->disks[disk].flags & AR_DF_ASSIGNED)
config->magic_0 = rdp->magic_0;
config->disk_number = disk;
switch (rdp->flags & (AR_F_RAID0 | AR_F_RAID1 | AR_F_SPAN)) {
case AR_F_RAID0:
config->type = HPT_T_RAID0;
if (rdp->disks[disk].flags & AR_DF_ONLINE)
config->order = HPT_O_READY;
else
config->order = HPT_O_DOWN;
break;
case AR_F_RAID1:
config->type = HPT_T_RAID1;
config->disk_number = (disk < rdp->width) ? disk : disk + 10;
break;
case AR_F_RAID0 | AR_F_RAID1:
config->magic_1 = rdp->magic_1;
config->type = HPT_T_RAID01_RAID0;
#if 0
if ((rdp->flags & (AR_F_READY | AR_F_DEGRADED)) == AR_F_READY)
if (disk < rdp->width)
config->order = HPT_O_RAID01SRC;
else
config->order = HPT_O_RAID01DST;
else
if (rdp->disks[disk].flags & AR_DF_ONLINE)
config->order = HPT_O_RAID01DEGRADED;
else
config->order = HPT_O_DOWN;
#else
if (rdp->disks[disk].flags & AR_DF_ONLINE)
if (disk < rdp->width)
config->order = HPT_O_RAID01SRC;
else
config->order = HPT_O_RAID01DST;
else
config->order = HPT_O_DOWN;
#endif
if (disk >= rdp->width) {
config->magic_0 = rdp->magic_0 + 1;
config->disk_number -= rdp->width;
}
break;
case AR_F_SPAN:
config->type = HPT_T_SPAN;
break;
}
config->array_width = rdp->width;
config->stripe_shift = (rdp->width > 1) ? (ffs(rdp->interleave)-1) : 0;
config->total_sectors = rdp->total_sectors;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
if ((rdp->disks[disk].flags & AR_DF_PRESENT) &&
rdp->disks[disk].device && rdp->disks[disk].device->driver &&
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
!(rdp->disks[disk].device->flags & ATA_D_DETACHING)) {
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
if (ar_rw(AD_SOFTC(rdp->disks[disk]), HPT_LBA,
sizeof(struct highpoint_raid_conf),
(caddr_t)config, AR_WRITE)) {
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
if (bootverbose)
printf("ar%d: Highpoint write conf failed\n", rdp->lun);
return -1;
}
}
}
return 0;
}
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
static int
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
ar_promise_read_conf(struct ad_softc *adp, struct ar_softc **raidp)
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
{
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
struct promise_raid_conf *info;
Clean up the raid code a bit, also allow disks on HPT controllers to be swapped around and still be put in the correct order in a raid.
2000-10-22 12:17:57 +00:00
struct ar_softc *raid;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
u_int32_t magic, cksum, *ckptr;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
int array, count, disk, retval = 0;
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
if (!(info = (struct promise_raid_conf *)
malloc(sizeof(struct promise_raid_conf), M_AR, M_NOWAIT | M_ZERO)))
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
return retval;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
if (ar_rw(adp, PR_LBA(adp), sizeof(struct promise_raid_conf),
(caddr_t)info, AR_READ | AR_WAIT)) {
Minor changes to the ATA RAID support code, remove some verbosity and put some under bootverbose..
2000-10-18 18:49:42 +00:00
if (bootverbose)
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
printf("ar: Promise read conf failed\n");
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
goto promise_out;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
/* check if this is a Promise RAID struct */
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
if (strncmp(info->promise_id, PR_MAGIC, sizeof(PR_MAGIC))) {
Minor changes to the ATA RAID support code, remove some verbosity and put some under bootverbose..
2000-10-18 18:49:42 +00:00
if (bootverbose)
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
printf("ar: Promise check1 failed\n");
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
goto promise_out;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
}
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
/* check if the checksum is OK */
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
for (cksum = 0, ckptr = (int32_t *)info, count = 0; count < 511; count++)
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
cksum += *ckptr++;
if (cksum != *ckptr) {
Minor changes to the ATA RAID support code, remove some verbosity and put some under bootverbose..
2000-10-18 18:49:42 +00:00
if (bootverbose)
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
printf("ar: Promise check2 failed\n");
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
goto promise_out;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
/* now convert Promise config info into our generic form */
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
if (info->raid.integrity != PR_I_VALID) {
Update the promise raid structure with some of the info I've gathered before I'm accused of "lending" it from somebody else.
2001-10-04 18:02:26 +00:00
if (bootverbose)
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
printf("ar: Promise check3 failed\n");
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
goto promise_out;
Update the config gathering code for both Promise & HPT
2000-11-01 17:35:44 +00:00
}
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
for (array = 0; array < MAX_ARRAYS; array++) {
if (!raidp[array]) {
raidp[array] =
(struct ar_softc*)malloc(sizeof(struct ar_softc), M_AR,
M_NOWAIT | M_ZERO);
if (!raidp[array]) {
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
printf("ar%d: failed to allocate raid config storage\n", array);
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
goto promise_out;
}
Update the config gathering code for both Promise & HPT
2000-11-01 17:35:44 +00:00
}
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
raid = raidp[array];
if (raid->flags & AR_F_HIGHPOINT_RAID)
continue;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
magic = (adp->device->channel->chiptype >> 16) |
(info->raid.array_number << 16);
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
if (raid->flags & AR_F_PROMISE_RAID && magic != raid->magic_0)
continue;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
/* update our knowledge about the array config based on generation */
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
if (!info->raid.generation || info->raid.generation > raid->generation){
raid->generation = info->raid.generation;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
raid->flags = AR_F_PROMISE_RAID;
raid->magic_0 = magic;
raid->lun = array;
if ((info->raid.status &
(PR_S_VALID | PR_S_ONLINE | PR_S_INITED | PR_S_READY)) ==
(PR_S_VALID | PR_S_ONLINE | PR_S_INITED | PR_S_READY)) {
raid->flags |= AR_F_READY;
if (info->raid.status & PR_S_DEGRADED)
raid->flags |= AR_F_DEGRADED;
}
else
raid->flags &= ~AR_F_READY;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
switch (info->raid.type) {
case PR_T_RAID0:
raid->flags |= AR_F_RAID0;
break;
case PR_T_RAID1:
raid->flags |= AR_F_RAID1;
if (info->raid.array_width > 1)
raid->flags |= AR_F_RAID0;
break;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
case PR_T_SPAN:
raid->flags |= AR_F_SPAN;
break;
default:
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
printf("ar%d: Promise unknown RAID type 0x%02x\n",
array, info->raid.type);
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
goto promise_out;
}
raid->interleave = 1 << info->raid.stripe_shift;
raid->width = info->raid.array_width;
raid->total_disks = info->raid.total_disks;
raid->heads = info->raid.heads + 1;
raid->sectors = info->raid.sectors;
raid->cylinders = info->raid.cylinders + 1;
raid->total_sectors = info->raid.total_sectors;
raid->offset = 0;
raid->reserved = 63;
/* convert disk flags to our internal types */
for (disk = 0; disk < info->raid.total_disks; disk++) {
raid->disks[disk].flags = 0;
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
if (info->raid.disk[disk].flags & PR_F_ONLINE)
raid->disks[disk].flags |= AR_DF_ONLINE;
if (info->raid.disk[disk].flags & PR_F_ASSIGNED)
raid->disks[disk].flags |= AR_DF_ASSIGNED;
if (info->raid.disk[disk].flags & PR_F_SPARE)
raid->disks[disk].flags |= AR_DF_SPARE;
if (info->raid.disk[disk].flags & (PR_F_REDIR | PR_F_DOWN)) {
raid->disks[disk].flags &= ~AR_DF_ONLINE;
raid->disks[disk].flags |= AR_DF_PRESENT;
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
}
}
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
if (raid->disks[info->raid.disk_number].flags && adp->device) {
raid->disks[info->raid.disk_number].device = adp->device;
raid->disks[info->raid.disk_number].flags |= AR_DF_PRESENT;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
raid->disks[info->raid.disk_number].disk_sectors =
info->raid.disk_sectors;
retval = 1;
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
}
Major update of the ATA RAID code, part 1: Overhaul of the attach/detach code and structures, there were some nasty bugs in the old implementation. This made it possible to collapse the ATA/ATAPI device control structures into one generic structure. A note here, the kernel is NOT ready for detach of active devices, it fails all over in random places, but for inactive devices it works. However for ATA RAID this works, since the RAID abstration layer insulates the buggy^H^H^H^H^H^Hfragile device subsystem from the physical disks. Proberly detect the RAID's from the BIOS, and mark critical RAID1 arrays as such, but continue if there is enough of the mirror left to do so. Properly fail arrays on a live system. For RAID0 that means return EIO, and for RAID1 it means continue on the still working part of the mirror if possible, else return EIO. If the state changes, log this to the console. Allow for Promise & Highpoint controllers/arrays to coexist on the same machine. It is not possible to distribute arrays over different makes of controllers though. If Promise SuperSwap enclosures are used, signal disk state on the status LED on the front. Misc fixes that I had lying around for various minor bugs. Sponsored by: Advanis Inc.
2002-02-04 19:23:40 +00:00
break;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
Overhaul to minimize stack usage, in some places >2K was used on the stack *blush*...
2001-09-20 15:25:36 +00:00
promise_out:
free(info, M_AR);
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
return retval;
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
}
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
static int
ar_promise_write_conf(struct ar_softc *rdp)
{
struct promise_raid_conf *config;
struct timeval timestamp;
u_int32_t *ckptr;
int count, disk, drive;
rdp->generation++;
microtime(&timestamp);
for (disk = 0; disk < rdp->total_disks; disk++) {
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
if (!(config = (struct promise_raid_conf *)
malloc(sizeof(struct promise_raid_conf), M_AR, M_NOWAIT))) {
printf("ar%d: Promise write conf failed\n", rdp->lun);
return -1;
}
for (count = 0; count < sizeof(struct promise_raid_conf); count++)
*(((u_int8_t *)config) + count) = 255 - (count % 256);
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
bcopy(PR_MAGIC, config->promise_id, sizeof(PR_MAGIC));
config->dummy_0 = 0x00020000;
config->magic_0 = PR_MAGIC0(rdp->disks[disk]) | timestamp.tv_sec;
config->magic_1 = timestamp.tv_sec >> 16;
config->magic_2 = timestamp.tv_sec;
config->raid.integrity = PR_I_VALID;
config->raid.flags = 0;
if (rdp->disks[disk].flags & AR_DF_PRESENT)
config->raid.flags |= PR_F_VALID;
if (rdp->disks[disk].flags & AR_DF_ASSIGNED)
config->raid.flags |= PR_F_ASSIGNED;
if (rdp->disks[disk].flags & AR_DF_ONLINE)
config->raid.flags |= PR_F_ONLINE;
else
config->raid.flags |= PR_F_DOWN;
config->raid.disk_number = disk;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
if (rdp->disks[disk].flags & AR_DF_PRESENT && rdp->disks[disk].device) {
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
config->raid.channel = rdp->disks[disk].device->channel->unit;
config->raid.device = (rdp->disks[disk].device->unit != 0);
if (AD_SOFTC(rdp->disks[disk])->dev->si_disk)
config->raid.disk_sectors = PR_LBA(AD_SOFTC(rdp->disks[disk]));
/*config->raid.disk_offset*/
}
config->raid.magic_0 = config->magic_0;
config->raid.rebuild_lba = 0xffffffff;
config->raid.generation = rdp->generation;
if (rdp->flags & AR_F_READY) {
config->raid.status =
(PR_S_VALID | PR_S_ONLINE | PR_S_INITED | PR_S_READY);
if (rdp->flags & AR_F_DEGRADED)
config->raid.status |= PR_S_DEGRADED;
else
config->raid.status |= PR_S_FUNCTIONAL;
}
else
config->raid.status = 0;
switch (rdp->flags & (AR_F_RAID0 | AR_F_RAID1 | AR_F_SPAN)) {
case AR_F_RAID0:
config->raid.type = PR_T_RAID0;
break;
case AR_F_RAID1:
config->raid.type = PR_T_RAID1;
break;
case AR_F_RAID0 | AR_F_RAID1:
config->raid.type = PR_T_RAID1;
break;
case AR_F_SPAN:
config->raid.type = PR_T_SPAN;
break;
}
config->raid.total_disks = rdp->total_disks;
config->raid.stripe_shift = ffs(rdp->interleave) - 1;
config->raid.array_width = rdp->width;
config->raid.array_number = rdp->magic_0 >> 16;
config->raid.total_sectors = rdp->total_sectors;
config->raid.cylinders = rdp->cylinders - 1;
config->raid.heads = rdp->heads - 1;
config->raid.sectors = rdp->sectors;
config->raid.magic_1 = (u_int64_t)config->magic_2<<16 | config->magic_1;
bzero(&config->raid.disk, 8 * 12);
for (drive = 0; drive < rdp->total_disks; drive++) {
config->raid.disk[drive].flags = 0;
if (rdp->disks[drive].flags & AR_DF_PRESENT)
config->raid.disk[drive].flags |= PR_F_VALID;
if (rdp->disks[drive].flags & AR_DF_ASSIGNED)
config->raid.disk[drive].flags |= PR_F_ASSIGNED;
if (rdp->disks[drive].flags & AR_DF_ONLINE)
config->raid.disk[drive].flags |= PR_F_ONLINE;
else
config->raid.disk[drive].flags = (PR_F_REDIR | PR_F_DOWN);
if (rdp->disks[drive].flags & AR_DF_SPARE)
config->raid.disk[drive].flags |= PR_F_SPARE;
config->raid.disk[drive].dummy_0 = 0x0;
if (rdp->disks[drive].device) {
config->raid.disk[drive].channel =
rdp->disks[drive].device->channel->unit;
config->raid.disk[drive].device =
(rdp->disks[drive].device->unit != 0);
}
config->raid.disk[drive].magic_0 =
PR_MAGIC0(rdp->disks[drive]) | timestamp.tv_sec;
}
config->checksum = 0;
for (ckptr = (int32_t *)config, count = 0; count < 511; count++)
config->checksum += *ckptr++;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
if ((rdp->disks[disk].flags & AR_DF_PRESENT) &&
rdp->disks[disk].device && rdp->disks[disk].device->driver &&
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
!(rdp->disks[disk].device->flags & ATA_D_DETACHING)) {
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
if (ar_rw(AD_SOFTC(rdp->disks[disk]),
PR_LBA(AD_SOFTC(rdp->disks[disk])),
sizeof(struct promise_raid_conf),
(caddr_t)config, AR_WRITE)) {
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
if (bootverbose)
printf("ar%d: Promise write conf failed\n", rdp->lun);
return -1;
}
}
}
return 0;
}
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
static void
ar_rw_done(struct bio *bp)
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
{
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
free(bp->bio_data, M_AR);
free(bp, M_AR);
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
}
static int
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
ar_rw(struct ad_softc *adp, u_int32_t lba, int count, caddr_t data, int flags)
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
{
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
struct bio *bp;
int s;
if (!(bp = (struct bio *)malloc(sizeof(struct bio), M_AR, M_NOWAIT|M_ZERO)))
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
return 1;
Major update of the ATA RAID code, part 3: Add code to properly detach/attach disks that are part of a RAID. Mark a disk that is attached on an ATA channel belonging to a RAID as a spare disk that can be used for rebuilding failed RAID1's. Add support for rebuilding failed RAID1's. Several fixes to the detach/attach code. For replacing a disk in a failed RAID1 do the following: Find the controller channel# of the failed disk. Exec 'atacontrol detach <channel#>' to free the disk from the system. Replace the failed disk with a new one of at least the same size. If your have your disks in drawers/enclosures this can be done with the system still running. Exec 'atacontrol attach <channel#>' to add the disk to the system and mark it as a valid spare for rebuild. Exec 'atacontrol rebuild <array#>' The system will rebuild the array on the fly, the array can still be used during this, although with slower performance. Please let me know of any problems with this! Sponsored by: Advanis Inc. MFC after: 2 weeks
2002-03-03 15:36:21 +00:00
bp->bio_dev = adp->dev;
bp->bio_data = data;
bp->bio_pblkno = lba;
bp->bio_bcount = count;
if (flags & AR_READ)
bp->bio_cmd = BIO_READ;
if (flags & AR_WRITE)
bp->bio_cmd = BIO_WRITE;
if (flags & AR_WAIT)
bp->bio_done = (void *)wakeup;
else
bp->bio_done = ar_rw_done;
s = splbio();
bp->bio_dev->AD_STRATEGY(bp);
splx(s);
if (flags & AR_WAIT) {
tsleep(bp, PRIBIO, "arrw", 0);
free(bp, M_AR);
Major update of the ATA RAID code, part 2: More cleanups of the RAID1 failure mode code. Add functionality that writes the changed RAID config setup back to the disks (in controller BIOS specific format), so that a reboot will make the BIOS pick up the changed config.
2002-02-12 11:35:15 +00:00
}
Add support for ATA "pseudo" RAID controllers as the Promise Fasttrak and HighPoint HPT370 controllers. Use by defining the RAID in the BIOS and the "ar driver will pick it up automagically...
2000-10-13 13:04:45 +00:00
return 0;
}
Reference in New Issue Copy Permalink