d/freebsd-nq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
c0da2d161e
freebsd-nq/sys/cam/scsi/scsi_da.c

4039 lines
105 KiB
C
Raw Normal View History

Start each of the license/copyright comments with /*-
2005-01-05 22:34:37 +00:00
/*-
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
* Implementation of SCSI Direct Access Peripheral driver for CAM.
*
* Copyright (c) 1997 Justin T. Gibbs.
* 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. 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 AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
* ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
Use __FBSDID().
2003-06-10 18:14:05 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
#include <sys/param.h>
Implement a new camcontrol function, 'camcontrol format'. libcam/Makefile: Add scsi_da.c to libcam for the new scsi_format_unit() function. camcontrol.8: Update the man page for the new format functionality, and take out the examples section describing how to do it with 'camcontrol cmd'. camcontrol.c: New format functionality. Note that unlike the rest of the camcontrol subcommands, this one is interactive by default. Because of the potential destructiveness of the format command, I thought it necessary to get confirmation from the user before spamming a disk. You can disable the interactive behavior, and the status meter with command line arguments. scsi_da.c: Add the new scsi_format_unit() cdb building function and use #ifdef _KERNEL to make this file compile in both the kernel and userland. The format unit function is currently only defined in the non-kernel case, because nothing in the kernel is using it. If that changes, it should be un-ifdefed and compiled in both cases. scsi_da.h: New function declaration, CDB structure and format data structures. Thanks to Nick Hibma for providing some valuable input on these changes.
2000-05-21 23:57:52 +00:00
#ifdef _KERNEL
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
#include <sys/systm.h>
#include <sys/kernel.h>
Separate the struct bio related stuff out of <sys/buf.h> into <sys/bio.h>. <sys/bio.h> is now a prerequisite for <sys/buf.h> but it shall not be made a nested include according to bdes teachings on the subject of nested includes. Diskdrivers and similar stuff below specfs::strategy() should no longer need to include <sys/buf.> unless they need caching of data. Still a few bogus uses of struct buf to track down. Repocopy by: peter
2000-05-05 09:59:14 +00:00
#include <sys/bio.h>
Add SYSCTL ints for default normal I/O timeout && retry counts. This is useful if you want to dynamically move into a Fibre Channel or Multi-initiator environment that happens to be particularly noisy and ugly that requires a lot of retries (with shorter I/O timeouts) for commands destried by LIPs or Bus Resets. Reviewed by: deafening silence on audit && scsi on the retry counts MFC after: 2 weeks
2001-07-09 19:18:00 +00:00
#include <sys/sysctl.h>
Unbreak buildworld. sys/taskqueue.h is a kernel-only include. Pointy Hat to: ken
2003-09-03 12:31:03 +00:00
#include <sys/taskqueue.h>
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
#include <sys/lock.h>
#include <sys/mutex.h>
Introduce the disk mini-layer and devstat_end_transaction_buf() in cam/scsi. Somewhat reviewed by: ken
1999-10-01 09:34:10 +00:00
#include <sys/conf.h>
Remove obsolete headers. Move kernel specific headers under #ifdefs
2007-04-23 18:01:44 +00:00
#include <sys/devicestat.h>
Implement a new generic mechanism for attaching handler functions to events, in order to pave the way for removing a number of the ad-hoc implementations currently in use. Retire the at_shutdown family of functions and replace them with new event handler lists. Rework kern_shutdown.c to take greater advantage of the use of event handlers. Reviewed by: green
1999-08-21 06:24:40 +00:00
#include <sys/eventhandler.h>
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
#include <sys/malloc.h>
Merge the cons.c and cons.h to the best of my ability. alpha may or may not compile, I can't test it.
1999-08-09 10:35:05 +00:00
#include <sys/cons.h>
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
#include <sys/endian.h>
Restore use of polling mode for disk cache flush in case of kernel panic. While I am not sure that any extra hardware access is a good idea after panic, that is an existing behaviour that should better work correctly.
2013-06-15 12:46:38 +00:00
#include <sys/proc.h>
Plumb device physical path reporting from CAM devices, through GEOM and DEVFS, and make it accessible via the diskinfo utility. Extend GEOM's generic attribute query mechanism into generic disk consumers. sys/geom/geom_disk.c: sys/geom/geom_disk.h: sys/cam/scsi/scsi_da.c: sys/cam/ata/ata_da.c: - Allow disk providers to implement a new method which can override the default BIO_GETATTR response, d_getattr(struct bio *). This function returns -1 if not handled, otherwise it returns 0 or an errno to be passed to g_io_deliver(). sys/cam/scsi/scsi_da.c: sys/cam/ata/ata_da.c: - Don't copy the serial number to dp->d_ident anymore, as the CAM XPT is now responsible for returning this information via d_getattr()->(a)dagetattr()->xpt_getatr(). sys/geom/geom_dev.c: - Implement a new ioctl, DIOCGPHYSPATH, which returns the GEOM attribute "GEOM::physpath", if possible. If the attribute request returns a zero-length string, ENOENT is returned. usr.sbin/diskinfo/diskinfo.c: - If the DIOCGPHYSPATH ioctl is successful, report physical path data when diskinfo is executed with the '-v' option. Submitted by: will Reviewed by: gibbs Sponsored by: Spectra Logic Corporation Add generic attribute change notification support to GEOM. sys/sys/geom/geom.h: Add a new attrchanged method field to both g_class and g_geom. sys/sys/geom/geom.h: sys/geom/geom_event.c: - Provide the g_attr_changed() function that providers can use to advertise attribute changes. - Perform delivery of attribute change notifications from a thread context via the standard GEOM event mechanism. sys/geom/geom_subr.c: Inherit the attrchanged method from class to geom (class instance). sys/geom/geom_disk.c: Provide disk_attr_changed() to provide g_attr_changed() access to consumers of the disk API. sys/cam/scsi/scsi_pass.c: sys/cam/scsi/scsi_da.c: sys/geom/geom_dev.c: sys/geom/geom_disk.c: Use attribute changed events to track updates to physical path information. sys/cam/scsi/scsi_da.c: Add AC_ADVINFO_CHANGED to the registered asynchronous CAM events for this driver. When this event occurs, and the updated buffer type references our physical path attribute, emit a GEOM attribute changed event via the disk_attr_changed() API. sys/cam/scsi/scsi_pass.c: Add AC_ADVINFO_CHANGED to the registered asynchronous CAM events for this driver. When this event occurs, update the physical patch devfs alias for this pass instance. Submitted by: gibbs Sponsored by: Spectra Logic Corporation
2011-06-14 17:10:32 +00:00
#include <geom/geom.h>
Use bioq_flush() to drain a bio queue with a specific error code. Retain the mistake of not updating the devstat API for now. Spell bioq_disksort() consistently with the remaining bioq_*(). #include <geom/geom_disk.h> where this is more appropriate.
2003-04-01 15:06:26 +00:00
#include <geom/geom_disk.h>
Remove obsolete headers. Move kernel specific headers under #ifdefs
2007-04-23 18:01:44 +00:00
#endif /* _KERNEL */
Use bioq_flush() to drain a bio queue with a specific error code. Retain the mistake of not updating the devstat API for now. Spell bioq_disksort() consistently with the remaining bioq_*(). #include <geom/geom_disk.h> where this is more appropriate.
2003-04-01 15:06:26 +00:00
Implement a new camcontrol function, 'camcontrol format'. libcam/Makefile: Add scsi_da.c to libcam for the new scsi_format_unit() function. camcontrol.8: Update the man page for the new format functionality, and take out the examples section describing how to do it with 'camcontrol cmd'. camcontrol.c: New format functionality. Note that unlike the rest of the camcontrol subcommands, this one is interactive by default. Because of the potential destructiveness of the format command, I thought it necessary to get confirmation from the user before spamming a disk. You can disable the interactive behavior, and the status meter with command line arguments. scsi_da.c: Add the new scsi_format_unit() cdb building function and use #ifdef _KERNEL to make this file compile in both the kernel and userland. The format unit function is currently only defined in the non-kernel case, because nothing in the kernel is using it. If that changes, it should be un-ifdefed and compiled in both cases. scsi_da.h: New function declaration, CDB structure and format data structures. Thanks to Nick Hibma for providing some valuable input on these changes.
2000-05-21 23:57:52 +00:00
#ifndef _KERNEL
#include <stdio.h>
#include <string.h>
#endif /* _KERNEL */
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
#include <cam/cam.h>
#include <cam/cam_ccb.h>
#include <cam/cam_periph.h>
#include <cam/cam_xpt_periph.h>
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
#include <cam/cam_sim.h>
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
#include <cam/scsi/scsi_message.h>
Implement a new camcontrol function, 'camcontrol format'. libcam/Makefile: Add scsi_da.c to libcam for the new scsi_format_unit() function. camcontrol.8: Update the man page for the new format functionality, and take out the examples section describing how to do it with 'camcontrol cmd'. camcontrol.c: New format functionality. Note that unlike the rest of the camcontrol subcommands, this one is interactive by default. Because of the potential destructiveness of the format command, I thought it necessary to get confirmation from the user before spamming a disk. You can disable the interactive behavior, and the status meter with command line arguments. scsi_da.c: Add the new scsi_format_unit() cdb building function and use #ifdef _KERNEL to make this file compile in both the kernel and userland. The format unit function is currently only defined in the non-kernel case, because nothing in the kernel is using it. If that changes, it should be un-ifdefed and compiled in both cases. scsi_da.h: New function declaration, CDB structure and format data structures. Thanks to Nick Hibma for providing some valuable input on these changes.
2000-05-21 23:57:52 +00:00
#ifndef _KERNEL
#include <cam/scsi/scsi_da.h>
#endif /* !_KERNEL */
#ifdef _KERNEL
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
typedef enum {
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
DA_STATE_PROBE_RC,
DA_STATE_PROBE_RC16,
DA_STATE_PROBE_LBP,
DA_STATE_PROBE_BLK_LIMITS,
Added automatic detection of non-rotating media which disables the use of BIO queue sorting, hence optimising performance for devices such as SSD's Reviewed by: scottl Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:31:03 +00:00
DA_STATE_PROBE_BDC,
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
DA_STATE_PROBE_ATA,
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
DA_STATE_NORMAL
} da_state;
typedef enum {
DA_FLAG_PACK_INVALID = 0x001,
DA_FLAG_NEW_PACK = 0x002,
DA_FLAG_PACK_LOCKED = 0x004,
DA_FLAG_PACK_REMOVABLE = 0x008,
DA_FLAG_NEED_OTAG = 0x020,
Some microoptimizations for da and ada drivers: - Replace ordered_tag_count counter with single flag; - From da remove outstanding_cmds counter, duplicating pending_ccbs list; - From da_softc remove unused links field.
2013-10-24 14:05:44 +00:00
DA_FLAG_WAS_OTAG = 0x040,
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
DA_FLAG_RETRY_UA = 0x080,
(dacleanup, dasysctlinit): Defend against calling sysctl_ctx_free on an uninitialized sysctl_ctx, using flag DA_FLAG_SCTX_INIT. This prevents a panic encoutered with some umass units that probe correctly but fail to attach. Same problem, and same fix, as scsi_cd.c rev. 1.86. Reviewed by: njl, ken
2003-10-08 07:12:30 +00:00
DA_FLAG_OPEN = 0x100,
Use READ CAPACITY(16) to get information about device physical sectors. As soon as not all devices support READ CAPACITY(16), automatically fall back to READ CAPACITY(10) if CAM_REQ_INVALID or SSD_KEY_ILLEGAL_REQUEST status returned. It also provides first bits of information about Logical Block Provisioning (aka UNMAP/TRIM) support by the device.
2011-12-23 19:12:02 +00:00
DA_FLAG_SCTX_INIT = 0x200,
Refactor da(4) to remove one of two code paths used to query capacity data. Reviewed by: ken, mav (earlier version) Sponsored by: The FreeBSD Foundation
2012-04-12 12:58:14 +00:00
DA_FLAG_CAN_RC16 = 0x400,
Synchronize device cache on close only if there were some write operations. While these operations are not really needed otherwise, at least for SCSI they may cause extra errors if some other initiator holds write exclusive reservation on the LUN (SYNCHRONIZE CACHE handled as "write" operation).
2013-07-27 22:44:55 +00:00
DA_FLAG_PROBED = 0x800,
Do not reread SCSI disk VPD pages on every device open. Instead of rereading VPD pages on every device open, do it only on initial device probe, and in cases when device reported via UNIT ATTENTIONs that something has changed. Capacity is still rereaded on every open because it is more critical for operation and more probable to change in run time. On my tests with Intel 530 SSDs on mps(4) HBA this change reduces time GEOM needs to retaste the device (that includes few open/close cycles) from ~150ms to ~30ms. MFC after: 2 weeks Sponsored by: iXsystems, Inc.
2014-04-30 17:38:26 +00:00
DA_FLAG_DIRTY = 0x1000,
DA_FLAG_ANNOUNCED = 0x2000
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
} da_flags;
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
typedef enum {
DA_Q_NONE = 0x00,
"Fix" a problem with the Quantum Viking. It appears that this drive does not like the 6-byte read and write commands! It returns illegal request, with the field pointer pointing to byte 9 of a 6 byte CDB. In any case, the work around is to put in a quirk mechanism that makes sure that we don't send 6-byte reads or writes to this device. It's rather sad that this is necessary. You'd think that they would be able to get something that basic to work right in their firmware... Reviewed by: gibbs Reported by: Adam McDougall <bsdx@spawnet.com>
1998-12-02 17:35:28 +00:00
DA_Q_NO_SYNC_CACHE = 0x01,
Add the DA_Q_NO_PREVENT quirk which keeps da(4) from sending PREVENT/ALLOW commands. Add a quirk for the Creative Nomad MuVo USB device that uses it as well as NO_SYNCHRONIZE_CACHE. PR: kern/53094 Submitted by: Richard Nyberg <rnyberg@it.su.se> MFC after: 3 days
2003-08-22 16:35:53 +00:00
DA_Q_NO_6_BYTE = 0x02,
Merge to da driver quirks hinting 4K physical sector sizes for SATA disks connected via SAS or USB. Unluckily I've found that SAS (mps) and USB-SATA I have translate models in different ways, requiring twice more quirks. Unluckily for Hitachi, their model names are trimmed on SAS, making impossible to identify 4K sector drives that way.
2011-12-23 00:10:17 +00:00
DA_Q_NO_PREVENT = 0x04,
Add NO_RC16 quirk to make da driver avoid using READ CAPACITY(16) command if possible. Use it for Kingston JetFlash USB sticks, that are known to return garbage in response to that command.
2013-07-30 13:00:09 +00:00
DA_Q_4K = 0x08,
Disable UNMAP support for STEC 842 SSDs. In some unknown cases UNMAP commands make device firmware stuck. MFC after: 2 weeks
2014-04-23 19:50:35 +00:00
DA_Q_NO_RC16 = 0x10,
Retry indefinitely on SCSI BUSY status from VMware disks and CDs. VMware returns BUSY status when storage has transient connectivity issues. It is often better to wait and let VM admin fix the problem then crash. Discussed with: ken MFC after: 1 week
2015-02-02 20:23:05 +00:00
DA_Q_NO_UNMAP = 0x20,
DA_Q_RETRY_BUSY = 0x40
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
} da_quirks;
Added output of device QUIRKS for CAM and AHCI devices during boot. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-05-18 23:36:21 +00:00
#define DA_Q_BIT_STRING \
"\020" \
"\001NO_SYNC_CACHE" \
"\002NO_6_BYTE" \
"\003NO_PREVENT" \
Add NO_RC16 quirk to make da driver avoid using READ CAPACITY(16) command if possible. Use it for Kingston JetFlash USB sticks, that are known to return garbage in response to that command.
2013-07-30 13:00:09 +00:00
"\0044K" \
Retry indefinitely on SCSI BUSY status from VMware disks and CDs. VMware returns BUSY status when storage has transient connectivity issues. It is often better to wait and let VM admin fix the problem then crash. Discussed with: ken MFC after: 1 week
2015-02-02 20:23:05 +00:00
"\005NO_RC16" \
"\006NO_UNMAP" \
"\007RETRY_BUSY"
Added output of device QUIRKS for CAM and AHCI devices during boot. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-05-18 23:36:21 +00:00
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
typedef enum {
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
DA_CCB_PROBE_RC = 0x01,
DA_CCB_PROBE_RC16 = 0x02,
DA_CCB_PROBE_LBP = 0x03,
DA_CCB_PROBE_BLK_LIMITS = 0x04,
Added automatic detection of non-rotating media which disables the use of BIO queue sorting, hence optimising performance for devices such as SSD's Reviewed by: scottl Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:31:03 +00:00
DA_CCB_PROBE_BDC = 0x05,
DA_CCB_PROBE_ATA = 0x06,
DA_CCB_BUFFER_IO = 0x07,
DA_CCB_DUMP = 0x0A,
DA_CCB_DELETE = 0x0B,
DA_CCB_TUR = 0x0C,
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
DA_CCB_TYPE_MASK = 0x0F,
DA_CCB_RETRY_UA = 0x10
} da_ccb_state;
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
/*
* Order here is important for method choice
*
* We prefer ATA_TRIM as tests run against a Sandforce 2281 SSD attached to
* LSI 2008 (mps) controller (FW: v12, Drv: v14) resulted 20% quicker deletes
* using ATA_TRIM than the corresponding UNMAP results for a real world mysql
* import taking 5mins.
*
*/
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
typedef enum {
DA_DELETE_NONE,
DA_DELETE_DISABLE,
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
DA_DELETE_ATA_TRIM,
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
DA_DELETE_UNMAP,
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
DA_DELETE_WS16,
DA_DELETE_WS10,
DA_DELETE_ZERO,
Enable CAM SCSI to choice ATA TRIM during autodetection and correct method names after increasing the priority of ATA TRIM. Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 15:03:30 +00:00
DA_DELETE_MIN = DA_DELETE_ATA_TRIM,
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
DA_DELETE_MAX = DA_DELETE_ZERO
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
} da_delete_methods;
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
typedef void da_delete_func_t (struct cam_periph *periph, union ccb *ccb,
struct bio *bp);
static da_delete_func_t da_delete_trim;
static da_delete_func_t da_delete_unmap;
static da_delete_func_t da_delete_ws;
Const-ify the new da_delete_functions. Remove a redundant sanity check Submitted by: Steven Hartland Obtained from: Netflix MFC after: 3 days
2013-07-12 23:20:11 +00:00
static const void * da_delete_functions[] = {
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
NULL,
NULL,
da_delete_trim,
da_delete_unmap,
da_delete_ws,
da_delete_ws,
da_delete_ws
};
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
static const char *da_delete_method_names[] =
Enable CAM SCSI to choice ATA TRIM during autodetection and correct method names after increasing the priority of ATA TRIM. Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 15:03:30 +00:00
{ "NONE", "DISABLE", "ATA_TRIM", "UNMAP", "WS16", "WS10", "ZERO" };
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
static const char *da_delete_method_desc[] =
Enable CAM SCSI to choice ATA TRIM during autodetection and correct method names after increasing the priority of ATA TRIM. Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 15:03:30 +00:00
{ "NONE", "DISABLED", "ATA TRIM", "UNMAP", "WRITE SAME(16) with UNMAP",
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
"WRITE SAME(10) with UNMAP", "ZERO" };
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
/* Offsets into our private area for storing information */
#define ccb_state ppriv_field0
#define ccb_bp ppriv_ptr1
struct disk_params {
u_int8_t heads;
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
u_int32_t cylinders;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
u_int8_t secs_per_track;
u_int32_t secsize; /* Number of bytes/sector */
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
u_int64_t sectors; /* total number sectors */
Merge to da driver quirks hinting 4K physical sector sizes for SATA disks connected via SAS or USB. Unluckily I've found that SAS (mps) and USB-SATA I have translate models in different ways, requiring twice more quirks. Unluckily for Hitachi, their model names are trimmed on SAS, making impossible to identify 4K sector drives that way.
2011-12-23 00:10:17 +00:00
u_int stripesize;
u_int stripeoffset;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
};
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
#define UNMAP_RANGE_MAX 0xffffffff
#define UNMAP_HEAD_SIZE 8
#define UNMAP_RANGE_SIZE 16
#define UNMAP_MAX_RANGES 2048 /* Protocol Max is 4095 */
#define UNMAP_BUF_SIZE ((UNMAP_MAX_RANGES * UNMAP_RANGE_SIZE) + \
UNMAP_HEAD_SIZE)
#define WS10_MAX_BLKS 0xffff
#define WS16_MAX_BLKS 0xffffffff
#define ATA_TRIM_MAX_RANGES ((UNMAP_BUF_SIZE / \
(ATA_DSM_RANGE_SIZE * ATA_DSM_BLK_SIZE)) * ATA_DSM_BLK_SIZE)
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
struct da_softc {
Complete the bio/buf divorce for all code below devfs::strategy Exceptions: Vinum untouched. This means that it cannot be compiled. Greg Lehey is on the case. CCD not converted yet, casts to struct buf (still safe) atapi-cd casts to struct buf to examine B_PHYS
2000-04-15 05:54:02 +00:00
struct bio_queue_head bio_queue;
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
struct bio_queue_head delete_queue;
struct bio_queue_head delete_run_queue;
Back out the previous change to the queue(3) interface. It was not discussed and should probably not happen. Requested by: msmith and others
2000-05-26 02:09:24 +00:00
LIST_HEAD(, ccb_hdr) pending_ccbs;
Some microoptimizations for da and ada drivers: - Replace ordered_tag_count counter with single flag; - From da remove outstanding_cmds counter, duplicating pending_ccbs list; - From da_softc remove unused links field.
2013-10-24 14:05:44 +00:00
int tur; /* TEST UNIT READY should be sent */
int refcount; /* Active xpt_action() calls */
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
da_state state;
da_flags flags;
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
da_quirks quirks;
Adds the ability to enable / disable sorting of BIO requests queued within CAM. This can significantly improve performance particularly for SSDs which don't suffer from seek latencies. The sysctl / tunable kern.cam.sort_io_queues provides the systems default setting where:- 0 = queued BIOs are NOT sorted 1 = queued BIOs are sorted (default) Each device gets its own sysctl kern.cam.<type>.<id>.sort_io_queue Valid values are:- -1 = use system default (default) 0 = queued BIOs are NOT sorted 1 = queued BIOs are sorted Note: Additional patch will look to add automatic use of none sorted queues for none rotating media e.g. SSD's Reviewed by: scottl Approved by: pjd (mentor) MFC after: 2 weeks
2013-03-29 22:58:15 +00:00
int sort_io_queue;
"Fix" a problem with the Quantum Viking. It appears that this drive does not like the 6-byte read and write commands! It returns illegal request, with the field pointer pointing to byte 9 of a 6 byte CDB. In any case, the work around is to put in a quirk mechanism that makes sure that we don't send 6-byte reads or writes to this device. It's rather sad that this is necessary. You'd think that they would be able to get something that basic to work right in their firmware... Reviewed by: gibbs Reported by: Adam McDougall <bsdx@spawnet.com>
1998-12-02 17:35:28 +00:00
int minimum_cmd_size;
Fix a race condition in CAM peripheral free handling, locking in the CAM XPT bus traversal code, and a number of other periph level issues. cam_periph.h, cam_periph.c: Modify cam_periph_acquire() to test the CAM_PERIPH_INVALID flag prior to allowing a reference count to be gained on a peripheral. Callers of this function will receive CAM_REQ_CMP_ERR status in the situation of attempting to reference an invalidated periph. This guarantees that a peripheral scheduled for a deferred free will not be accessed during its wait for destruction. Panic during attempts to drop a reference count on a peripheral that already has a zero reference count. In cam_periph_list(), use a local sbuf with SBUF_FIXEDLEN set so that mallocs do not occur while the xpt topology lock is held, regardless of the allocation policy of the passed in sbuf. Add a new routine, cam_periph_release_locked_buses(), that can be called when the caller already holds the CAM topology lock. Add some extra debugging for duplicate peripheral allocations in cam_periph_alloc(). Treat CAM_DEV_NOT_THERE much the same as a selection timeout (AC_LOST_DEVICE is emitted), but forgo retries. cam_xpt.c: Revamp the way the EDT traversal code does locking and reference counting. This was broken, since it assumed that the EDT would not change during traversal, but that assumption is no longer valid. So, to prevent devices from going away while we traverse the EDT, make sure we properly lock everything and hold references on devices that we are using. The two peripheral driver traversal routines should be examined. xptpdperiphtraverse() holds the topology lock for the entire time it runs. xptperiphtraverse() is now locked properly, but only holds the topology lock while it is traversing the list, and not while the traversal function is running. The bus locking code in xptbustraverse() should also be revisited at a later time, since it is complex and should probably be simplified. scsi_da.c: Pay attention to the return value from cam_periph_acquire(). Return 0 always from daclose() even if the disk is now gone. Add some rudimentary error injection support. scsi_sg.c: Fix reference counting in the sg(4) driver. The sg driver was calling cam_periph_release() on close, but never called cam_periph_acquire() (which increments the reference count) on open. The periph code correctly complained that the sg(4) driver was trying to decrement the refcount when it was already 0. Sponsored by: Spectra Logic MFC after: 2 weeks
2012-01-12 00:41:48 +00:00
int error_inject;
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
int trim_max_ranges;
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
int delete_running;
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
int delete_available; /* Delete methods possibly available */
Respect MAXIMUM TRANSFER LENGTH field of Block Limits VPD page. Nobody yet reported disk supporting I/Os less then our MAXPHYS value, but since we any way have code to read Block Limits VPD page, that is easy. MFC after: 2 weeks
2014-04-30 19:44:31 +00:00
u_int maxio;
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
uint32_t unmap_max_ranges;
Mostly revert r260267 and hopefully really fix the original problem. The latest draft of SBC-3 tells: "A MAXIMUM UNMAP LBA COUNT field set to a non-zero value indicates the maximum number of LBAs that may be unmapped by an UNMAP command." To me it does not sound like that limit is set per single descriptor, but rather per all command. And I have at least one device that behaves exactly that way. This patch fixes the problem there. MFC after: 1 week
2014-01-22 22:19:53 +00:00
uint32_t unmap_max_lba; /* Max LBAs in UNMAP req */
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
uint64_t ws_max_blks;
Make delete method set via kern.cam.da.X.delete_method persistent. This allows to set delete method via tunable, before device capabilities are known. Also allow ZERO method for devices not reporting LBP, if user explicitly requests it -- it may be useful if storage supports compression and WRITE SAME, but does not support UNMAP. MFC after: 2 weeks
2015-10-11 18:26:06 +00:00
da_delete_methods delete_method_pref;
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
da_delete_methods delete_method;
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
da_delete_func_t *delete_func;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
struct disk_params params;
Change the disk(9) API in order to make device removal more robust. Previously the "struct disk" were owned by the device driver and this gave us problems when the device disappared and the users of that device were not immediately disappearing. Now the struct disk is allocate with a new call, disk_alloc() and owned by geom_disk and just abandonned by the device driver when disk_create() is called. Unfortunately, this results in a ton of "s/\./->/" changes to device drivers. Since I'm doing the sweep anyway, a couple of other API improvements have been carried out at the same time: The Giant awareness flag has been flipped from DISKFLAG_NOGIANT to DISKFLAG_NEEDSGIANT A version number have been added to disk_create() so that we can detect, report and ignore binary drivers with old ABI in the future. Manual page update to follow shortly.
2004-02-18 21:36:53 +00:00
struct disk *disk;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
union ccb saved_ccb;
Move dynamic sysctl(8) variable creation for the cd(4) and da(4) drivers out of cdregister() and daregister(), which are run from interrupt context. The sysctl code does blocking mallocs (M_WAITOK), which causes problems if malloc(9) actually needs to sleep. The eventual fix for this issue will involve moving the CAM probe process inside a kernel thread. For now, though, I have fixed the issue by moving dynamic sysctl variable creation for these two drivers to a task queue running in a kernel thread. The existing task queues (taskqueue_swi and taskqueue_swi_giant) run in software interrupt handlers, which wouldn't fix the problem at hand. So I have created a new task queue, taskqueue_thread, that runs inside a kernel thread. (It also runs outside of Giant -- clients must explicitly acquire and release Giant in their taskqueue functions.) scsi_cd.c: Remove sysctl variable creation code from cdregister(), and move it to a new function, cdsysctlinit(). Queue cdsysctlinit() to the taskqueue_thread taskqueue once we have fully registered the cd(4) driver instance. scsi_da.c: Remove sysctl variable creation code from daregister(), and move it to move it to a new function, dasysctlinit(). Queue dasysctlinit() to the taskqueue_thread taskqueue once we have fully registered the da(4) instance. taskqueue.h: Declare the new taskqueue_thread taskqueue, update some comments. subr_taskqueue.c: Create the new kernel thread taskqueue. This taskqueue runs outside of Giant, so any functions queued to it would need to explicitly acquire/release Giant if they need it. cd.4: Update the cd(4) man page to talk about the minimum command size sysctl/loader tunable. Also note that the changer variables are available as loader tunables as well. da.4: Update the da(4) man page to cover the retry_count, default_timeout and minimum_cmd_size sysctl variables/loader tunables. Remove references to /dev/r???, they aren't used any longer. cd.9: Update the cd(9) man page to describe the CD_Q_10_BYTE_ONLY quirk. taskqueue.9: Update the taskqueue(9) man page to describe the new thread task queue, and the taskqueue_swi_giant queue. MFC after: 3 days
2003-09-03 04:46:28 +00:00
struct task sysctl_task;
Fix ATAPI/USB/Firewire CDROM drive handling in cd(4) and hopefully fix a number of related problems along the way. - Automatically detect CDROM drives that can't handle 6 byte mode sense and mode select, and adjust our command size accordingly. We have to handle this in the cd(4) driver (where the buffers are allocated), since the parameter list length is different for the 6 and 10 byte mode sense commands. - Remove MODE_SENSE and MODE_SELECT translation removed in ATAPICAM and in the umass(4) driver, since there's no way for that to work properly. - Add a quirk entry for CDROM drives that just hang when they get a 6 byte mode sense or mode select. The reason for the quirk must be documented in a PR, and all quirks must be approved by ken@FreeBSD.org. This is to make sure that we fully understand why each quirk is needed. Once the CAM_NEW_TRAN_CODE is finished, we should be able to remove any such quirks, since we'll know what protocol the drive speaks (SCSI, ATAPI, etc.) and therefore whether we should use 6 or 10 byte mode sense/select commands. - Change the way the da(4) handles the no_6_byte sysctl. There is now a per-drive sysctl to set the minimum command size for that particular disk. (Since you could have multiple disks with multiple requirements in one system.) - Loader tunable support for all the sysctls in the da(4) and cd(4) drivers. - Add a CDIOCCLOSE ioctl for cd(4) (bde pointed this out a long time ago). - Add a media validation routine (cdcheckmedia()) to the cd(4) driver, to fix some problems bde pointed out a long time ago. We now allow open() to succeed no matter what, but if we don't detect valid media, the user can only issue CDIOCCLOSE or CDIOCEJECT ioctls. - The media validation routine also reads the table of contents off the drive. We use the table of contents to implement the CDIOCPLAYTRACKS ioctl using the PLAY AUDIO MSF command. The PLAY AUDIO TRACK INDEX command that we previously used was deprecated after SCSI-2. It works in every SCSI CDROM I've tried, but doesn't seem to work on ATAPI CDROM drives. We still use the play audio track index command if we don't have a valid TOC, but I suppose it'll fail anyway in that case. - Add _len() versions of scsi_mode_sense() and scsi_mode_select() so that we can specify the minimum command length. - Fix a couple of formatting problems in the sense printing code. MFC after: 4 weeks
2003-02-21 06:19:38 +00:00
struct sysctl_ctx_list sysctl_ctx;
struct sysctl_oid *sysctl_tree;
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
struct callout sendordered_c;
Do a minor amount of stylifying. Also, get a Fibre Channel WWPN if one exists for a da unit and create a sysctl OID for it.
2010-06-07 17:41:34 +00:00
uint64_t wwpn;
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
uint8_t unmap_buf[UNMAP_BUF_SIZE];
Add CAM infrastructure to allow reporting when a drive's long read capacity data changes. cam_ccb.h: Add a new advanced information type, CDAI_TYPE_RCAPLONG, for long read capacity data. cam_xpt_internal.h: Add a read capacity data pointer and length to struct cam_ed. cam_xpt.c: Free the read capacity buffer when a device goes away. While we're here, make sure we don't leak memory for other malloced fields in struct cam_ed. scsi_all.c: Update the scsi_read_capacity_16() to take a uint8_t * and a length instead of just a pointer to the parameter data structure. This will hopefully make this function somewhat immune to future changes in the parameter data. scsi_all.h: Add some extra bit definitions to struct scsi_read_capacity_data_long, and bump up the structure size to the full size specified by SBC-3. Change the prototype for scsi_read_capacity_16(). scsi_da.c: Register changes in read capacity data with the transport layer. This allows the transport layer to send out an async notification to interested parties. Update the dasetgeom() API. Use scsi_extract_sense_len() instead of scsi_extract_sense(). scsi_xpt.c: Add support for the new CDAI_TYPE_RCAPLONG advanced information type. Make sure we set the physpath pointer to NULL after freeing it. This allows blindly freeing it in the struct cam_ed destructor. sys/param.h: Bump __FreeBSD_version from 1000005 to 1000006 to make it easier for third party drivers to determine that the read capacity data async notification is available. camcontrol.c, mptutil/mpt_cam.c: Update these for the new scsi_read_capacity_16() argument structure. Sponsored by: Spectra Logic
2012-01-26 18:09:28 +00:00
struct scsi_read_capacity_data_long rcaplong;
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
struct callout mediapoll_c;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
};
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
#define dadeleteflag(softc, delete_method, enable) \
if (enable) { \
softc->delete_available |= (1 << delete_method); \
} else { \
softc->delete_available &= ~(1 << delete_method); \
}
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
struct da_quirk_entry {
struct scsi_inquiry_pattern inq_pat;
da_quirks quirks;
};
Add quirk entries from Andre Albsmeier to disable the sync cache command for the Quantum "MAVERICK 540S" and "LPS525S". Also, add common string variables, since we seem to have a few Quantum and Micropolis drives in here. Fix the 'quantum' variable usage in scsi_all.c that likely got broken when someone staticized things in cam_xpt.c. (That particular problem would cause Quantum Fireball ST drives to not get spun up if they were not already spinning.) Submitted by: Andre Albsmeier <andre.albsmeier@mchp.siemens.de>
2000-08-29 22:09:23 +00:00
static const char quantum[] = "QUANTUM";
static const char microp[] = "MICROP";
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
static struct da_quirk_entry da_quirk_table[] =
{
Sort quirks into sections.
2003-08-25 18:14:43 +00:00
/* SPI, FC devices */
Add quirk for Fujitsu M2513A MO drives. These drives hang at various operations due to the synchronize cache command. PR: kern/21674 Submitted by: W.Scholten <whs@xs4all.nl> MFC after: 1 week
2001-11-17 14:46:22 +00:00
{
/*
* Fujitsu M2513A MO drives.
* Tested devices: M2513A2 firmware versions 1200 & 1300.
* (dip switch selects whether T_DIRECT or T_OPTICAL device)
* Reported by: W.Scholten <whs@xs4all.nl>
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "FUJITSU", "M2513A", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE
},
{
/* See above. */
{T_OPTICAL, SIP_MEDIA_REMOVABLE, "FUJITSU", "M2513A", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE
},
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
{
Add quirk entries to disable the synchronize cache command for Micropolis 2217's (reported by Matthew Jacob in NetBSD PR kern/6027) and Fujitsu M2954's (reported by Tom Jackson). Some of the Fujitsus at least hang when they get a cache sync command. (Others just return illegal request.) Also, make error printing in dashutdown() a little more selective. Don't print any error when the sense key is illegal request. Drives that don't support the synchronize cache command usually return illegal request. Also, make sure the scsi status is check condition before going into scsi_sense_print(). Reviewed by: gibbs
1998-10-12 17:16:47 +00:00
/*
* This particular Fujitsu drive doesn't like the
* synchronize cache command.
* Reported by: Tom Jackson <toj@gorilla.net>
*/
{T_DIRECT, SIP_MEDIA_FIXED, "FUJITSU", "M2954*", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE
},
{
/*
* This drive doesn't like the synchronize cache command
* either. Reported by: Matthew Jacob <mjacob@feral.com>
* in NetBSD PR kern/6027, August 24, 1998.
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
*/
Add quirk entries from Andre Albsmeier to disable the sync cache command for the Quantum "MAVERICK 540S" and "LPS525S". Also, add common string variables, since we seem to have a few Quantum and Micropolis drives in here. Fix the 'quantum' variable usage in scsi_all.c that likely got broken when someone staticized things in cam_xpt.c. (That particular problem would cause Quantum Fireball ST drives to not get spun up if they were not already spinning.) Submitted by: Andre Albsmeier <andre.albsmeier@mchp.siemens.de>
2000-08-29 22:09:23 +00:00
{T_DIRECT, SIP_MEDIA_FIXED, microp, "2217*", "*"},
Add quirk entries to disable the synchronize cache command for Micropolis 2217's (reported by Matthew Jacob in NetBSD PR kern/6027) and Fujitsu M2954's (reported by Tom Jackson). Some of the Fujitsus at least hang when they get a cache sync command. (Others just return illegal request.) Also, make error printing in dashutdown() a little more selective. Don't print any error when the sense key is illegal request. Drives that don't support the synchronize cache command usually return illegal request. Also, make sure the scsi status is check condition before going into scsi_sense_print(). Reviewed by: gibbs
1998-10-12 17:16:47 +00:00
/*quirks*/ DA_Q_NO_SYNC_CACHE
Disable cache syncs for a broken NEC drive. Reviewed by: gibbs Submitted by: Blaz Zupan <blaz@gold.amis.net>
1998-10-13 23:34:54 +00:00
},
Add a quirk that disables SYNCHRONIZE CACHE PR: 8882 Obtained from: Hellmuth Michaelis hm@kts.org
1999-01-05 20:43:41 +00:00
{
/*
* This drive doesn't like the synchronize cache command
* either. Reported by: Hellmuth Michaelis (hm@kts.org)
* (PR 8882).
*/
Add quirk entries from Andre Albsmeier to disable the sync cache command for the Quantum "MAVERICK 540S" and "LPS525S". Also, add common string variables, since we seem to have a few Quantum and Micropolis drives in here. Fix the 'quantum' variable usage in scsi_all.c that likely got broken when someone staticized things in cam_xpt.c. (That particular problem would cause Quantum Fireball ST drives to not get spun up if they were not already spinning.) Submitted by: Andre Albsmeier <andre.albsmeier@mchp.siemens.de>
2000-08-29 22:09:23 +00:00
{T_DIRECT, SIP_MEDIA_FIXED, microp, "2112*", "*"},
Add a quirk that disables SYNCHRONIZE CACHE PR: 8882 Obtained from: Hellmuth Michaelis hm@kts.org
1999-01-05 20:43:41 +00:00
/*quirks*/ DA_Q_NO_SYNC_CACHE
},
Disable cache syncs for a broken NEC drive. Reviewed by: gibbs Submitted by: Blaz Zupan <blaz@gold.amis.net>
1998-10-13 23:34:54 +00:00
{
/*
* Doesn't like the synchronize cache command.
* Reported by: Blaz Zupan <blaz@gold.amis.net>
*/
{T_DIRECT, SIP_MEDIA_FIXED, "NEC", "D3847*", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE
"Fix" a problem with the Quantum Viking. It appears that this drive does not like the 6-byte read and write commands! It returns illegal request, with the field pointer pointing to byte 9 of a 6 byte CDB. In any case, the work around is to put in a quirk mechanism that makes sure that we don't send 6-byte reads or writes to this device. It's rather sad that this is necessary. You'd think that they would be able to get something that basic to work right in their firmware... Reviewed by: gibbs Reported by: Adam McDougall <bsdx@spawnet.com>
1998-12-02 17:35:28 +00:00
},
Add quirk entries from Andre Albsmeier to disable the sync cache command for the Quantum "MAVERICK 540S" and "LPS525S". Also, add common string variables, since we seem to have a few Quantum and Micropolis drives in here. Fix the 'quantum' variable usage in scsi_all.c that likely got broken when someone staticized things in cam_xpt.c. (That particular problem would cause Quantum Fireball ST drives to not get spun up if they were not already spinning.) Submitted by: Andre Albsmeier <andre.albsmeier@mchp.siemens.de>
2000-08-29 22:09:23 +00:00
{
/*
* Doesn't like the synchronize cache command.
Take constructive advice from njl && reformat previously added quirks slightly.
2005-04-14 04:46:46 +00:00
* Reported by: Blaz Zupan <blaz@gold.amis.net>
Add quirk entries from Andre Albsmeier to disable the sync cache command for the Quantum "MAVERICK 540S" and "LPS525S". Also, add common string variables, since we seem to have a few Quantum and Micropolis drives in here. Fix the 'quantum' variable usage in scsi_all.c that likely got broken when someone staticized things in cam_xpt.c. (That particular problem would cause Quantum Fireball ST drives to not get spun up if they were not already spinning.) Submitted by: Andre Albsmeier <andre.albsmeier@mchp.siemens.de>
2000-08-29 22:09:23 +00:00
*/
{T_DIRECT, SIP_MEDIA_FIXED, quantum, "MAVERICK 540S", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE
},
{
/*
* Doesn't like the synchronize cache command.
*/
{T_DIRECT, SIP_MEDIA_FIXED, quantum, "LPS525S", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE
},
Applied conservative version of suggested quirk. PR: 57468 Submitted by: walter@pelissero.de MFC after: 1 week
2005-04-14 04:05:00 +00:00
{
/*
* Doesn't like the synchronize cache command.
Take constructive advice from njl && reformat previously added quirks slightly.
2005-04-14 04:46:46 +00:00
* Reported by: walter@pelissero.de
Applied conservative version of suggested quirk. PR: 57468 Submitted by: walter@pelissero.de MFC after: 1 week
2005-04-14 04:05:00 +00:00
*/
{T_DIRECT, SIP_MEDIA_FIXED, quantum, "LPS540S", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE
},
"Fix" a problem with the Quantum Viking. It appears that this drive does not like the 6-byte read and write commands! It returns illegal request, with the field pointer pointing to byte 9 of a 6 byte CDB. In any case, the work around is to put in a quirk mechanism that makes sure that we don't send 6-byte reads or writes to this device. It's rather sad that this is necessary. You'd think that they would be able to get something that basic to work right in their firmware... Reviewed by: gibbs Reported by: Adam McDougall <bsdx@spawnet.com>
1998-12-02 17:35:28 +00:00
{
/*
* Doesn't work correctly with 6 byte reads/writes.
* Returns illegal request, and points to byte 9 of the
* 6-byte CDB.
* Reported by: Adam McDougall <bsdx@spawnet.com>
*/
Add quirk entries from Andre Albsmeier to disable the sync cache command for the Quantum "MAVERICK 540S" and "LPS525S". Also, add common string variables, since we seem to have a few Quantum and Micropolis drives in here. Fix the 'quantum' variable usage in scsi_all.c that likely got broken when someone staticized things in cam_xpt.c. (That particular problem would cause Quantum Fireball ST drives to not get spun up if they were not already spinning.) Submitted by: Andre Albsmeier <andre.albsmeier@mchp.siemens.de>
2000-08-29 22:09:23 +00:00
{T_DIRECT, SIP_MEDIA_FIXED, quantum, "VIKING 4*", "*"},
"Fix" a problem with the Quantum Viking. It appears that this drive does not like the 6-byte read and write commands! It returns illegal request, with the field pointer pointing to byte 9 of a 6 byte CDB. In any case, the work around is to put in a quirk mechanism that makes sure that we don't send 6-byte reads or writes to this device. It's rather sad that this is necessary. You'd think that they would be able to get something that basic to work right in their firmware... Reviewed by: gibbs Reported by: Adam McDougall <bsdx@spawnet.com>
1998-12-02 17:35:28 +00:00
/*quirks*/ DA_Q_NO_6_BYTE
},
{
Add the DA_Q_NO_PREVENT quirk which keeps da(4) from sending PREVENT/ALLOW commands. Add a quirk for the Creative Nomad MuVo USB device that uses it as well as NO_SYNCHRONIZE_CACHE. PR: kern/53094 Submitted by: Richard Nyberg <rnyberg@it.su.se> MFC after: 3 days
2003-08-22 16:35:53 +00:00
/* See above. */
Add quirk entries from Andre Albsmeier to disable the sync cache command for the Quantum "MAVERICK 540S" and "LPS525S". Also, add common string variables, since we seem to have a few Quantum and Micropolis drives in here. Fix the 'quantum' variable usage in scsi_all.c that likely got broken when someone staticized things in cam_xpt.c. (That particular problem would cause Quantum Fireball ST drives to not get spun up if they were not already spinning.) Submitted by: Andre Albsmeier <andre.albsmeier@mchp.siemens.de>
2000-08-29 22:09:23 +00:00
{T_DIRECT, SIP_MEDIA_FIXED, quantum, "VIKING 2*", "*"},
"Fix" a problem with the Quantum Viking. It appears that this drive does not like the 6-byte read and write commands! It returns illegal request, with the field pointer pointing to byte 9 of a 6 byte CDB. In any case, the work around is to put in a quirk mechanism that makes sure that we don't send 6-byte reads or writes to this device. It's rather sad that this is necessary. You'd think that they would be able to get something that basic to work right in their firmware... Reviewed by: gibbs Reported by: Adam McDougall <bsdx@spawnet.com>
1998-12-02 17:35:28 +00:00
/*quirks*/ DA_Q_NO_6_BYTE
Add a quirk entry for Y-E Data USB floppy drive. Driver follows in the next few days.
2000-03-15 22:44:03 +00:00
},
Apply quirk. PR: 57469 Submitted by: walter@pelissero.de MFC after: 1 week
2005-04-14 04:12:45 +00:00
{
Take constructive advice from njl && reformat previously added quirks slightly.
2005-04-14 04:46:46 +00:00
/*
* Doesn't like the synchronize cache command.
* Reported by: walter@pelissero.de
*/
Apply quirk. PR: 57469 Submitted by: walter@pelissero.de MFC after: 1 week
2005-04-14 04:12:45 +00:00
{T_DIRECT, SIP_MEDIA_FIXED, "CONNER", "CP3500*", "*"},
o Style(9): trim eol w/spaces, w/spaces to tabs, kill an empty line.
2006-11-26 17:15:54 +00:00
/*quirks*/ DA_Q_NO_SYNC_CACHE
Apply quirk. PR: 57469 Submitted by: walter@pelissero.de MFC after: 1 week
2005-04-14 04:12:45 +00:00
},
Move the ciss quirk to the right section, also update the comment stating that the controllers do not support SYNC CACHE since ciss only supports a small subset of the scsi spec.
2003-12-08 06:29:38 +00:00
{
/*
* The CISS RAID controllers do not support SYNC_CACHE
*/
{T_DIRECT, SIP_MEDIA_FIXED, "COMPAQ", "RAID*", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE
},
Disable UNMAP support for STEC 842 SSDs. In some unknown cases UNMAP commands make device firmware stuck. MFC after: 2 weeks
2014-04-23 19:50:35 +00:00
{
/*
Extend UNMAP blacklist on all STEC SSD models. None of existing STEC devices need UNMAP or even support it well, having many limitations and even hanging sometimes executing those commands. New devices that may use UNMAP going to be released under HGST name. MFC after: 3 days
2014-09-10 21:24:15 +00:00
* The STEC SSDs sometimes hang on UNMAP.
Disable UNMAP support for STEC 842 SSDs. In some unknown cases UNMAP commands make device firmware stuck. MFC after: 2 weeks
2014-04-23 19:50:35 +00:00
*/
Extend UNMAP blacklist on all STEC SSD models. None of existing STEC devices need UNMAP or even support it well, having many limitations and even hanging sometimes executing those commands. New devices that may use UNMAP going to be released under HGST name. MFC after: 3 days
2014-09-10 21:24:15 +00:00
{T_DIRECT, SIP_MEDIA_FIXED, "STEC", "*", "*"},
Disable UNMAP support for STEC 842 SSDs. In some unknown cases UNMAP commands make device firmware stuck. MFC after: 2 weeks
2014-04-23 19:50:35 +00:00
/*quirks*/ DA_Q_NO_UNMAP
},
Retry indefinitely on SCSI BUSY status from VMware disks and CDs. VMware returns BUSY status when storage has transient connectivity issues. It is often better to wait and let VM admin fix the problem then crash. Discussed with: ken MFC after: 1 week
2015-02-02 20:23:05 +00:00
{
/*
* VMware returns BUSY status when storage has transient
* connectivity problems, so better wait.
*/
Extend mask of VMware virtual disks.
2015-11-05 09:07:53 +00:00
{T_DIRECT, SIP_MEDIA_FIXED, "VMware*", "*", "*"},
Retry indefinitely on SCSI BUSY status from VMware disks and CDs. VMware returns BUSY status when storage has transient connectivity issues. It is often better to wait and let VM admin fix the problem then crash. Discussed with: ken MFC after: 1 week
2015-02-02 20:23:05 +00:00
/*quirks*/ DA_Q_RETRY_BUSY
},
Sort quirks into sections.
2003-08-25 18:14:43 +00:00
/* USB mass storage devices supported by umass(4) */
Add quirks for the EXATELECOM i-Bead mp3 player. PR: kern/51675 Submitted by: Nicolas Jombart <ecu@ipv42.net> MFC after: 3 days
2003-08-22 05:43:30 +00:00
{
/*
* EXATELECOM (Sigmatel) i-Bead 100/105 USB Flash MP3 Player
* PR: kern/51675
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "EXATEL", "i-BEAD10*", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE
},
Quirk for Jungsoft NEXDISK USB flash key. Fails to mount without NO_SYNCHRONIZE_CACHE. PR: kern/54737 Submitted by: David Thiel <lx@redundancy.redundancy.org> MFC after: 3 days
2003-08-22 05:58:23 +00:00
{
/*
Add miscellaneous USB device quirks. PR: kern/53067 PR: kern/54737 PR: kern/54786 PR: kern/57046 MFC after: 1 day
2004-04-19 04:14:09 +00:00
* Power Quotient Int. (PQI) USB flash key
* PR: kern/53067
Quirk for Jungsoft NEXDISK USB flash key. Fails to mount without NO_SYNCHRONIZE_CACHE. PR: kern/54737 Submitted by: David Thiel <lx@redundancy.redundancy.org> MFC after: 3 days
2003-08-22 05:58:23 +00:00
*/
Add miscellaneous USB device quirks. PR: kern/53067 PR: kern/54737 PR: kern/54786 PR: kern/57046 MFC after: 1 day
2004-04-19 04:14:09 +00:00
{T_DIRECT, SIP_MEDIA_REMOVABLE, "Generic*", "USB Flash Disk*",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
o Style(9): trim eol w/spaces, w/spaces to tabs, kill an empty line.
2006-11-26 17:15:54 +00:00
},
Add the DA_Q_NO_PREVENT quirk which keeps da(4) from sending PREVENT/ALLOW commands. Add a quirk for the Creative Nomad MuVo USB device that uses it as well as NO_SYNCHRONIZE_CACHE. PR: kern/53094 Submitted by: Richard Nyberg <rnyberg@it.su.se> MFC after: 3 days
2003-08-22 16:35:53 +00:00
{
/*
* Creative Nomad MUVO mp3 player (USB)
* PR: kern/53094
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "CREATIVE", "NOMAD_MUVO", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE|DA_Q_NO_PREVENT
},
Add miscellaneous USB device quirks. PR: kern/53067 PR: kern/54737 PR: kern/54786 PR: kern/57046 MFC after: 1 day
2004-04-19 04:14:09 +00:00
{
/*
* Jungsoft NEXDISK USB flash key
* PR: kern/54737
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "JUNGSOFT", "NEXDISK*", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE
},
{
/*
* FreeDik USB Mini Data Drive
* PR: kern/54786
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "FreeDik*", "Mini Data Drive",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
{
/*
* Sigmatel USB Flash MP3 Player
* PR: kern/57046
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "SigmaTel", "MSCN", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE|DA_Q_NO_PREVENT
},
Quirk for Neuros USB audio device. PR: kern/63645 Submitted by: Aron Stansvik <elvstone@osdever.net>
2004-05-02 19:24:23 +00:00
{
/*
* Neuros USB Digital Audio Computer
* PR: kern/63645
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "NEUROS", "dig. audio comp.",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
Quirk for SEGRAND NP-900 USB MP3Player PR: kern/64563 Submitted by: Kunitada Kokubun <unix_grandy@yahoo.co.jp> MFC after: 1 week
2004-07-18 05:39:14 +00:00
{
/*
* SEAGRAND NP-900 MP3 Player
* PR: kern/64563
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "SEAGRAND", "NP-900*", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE|DA_Q_NO_PREVENT
},
Add support iRiver iFP MP3 player PR: kern/54881, i386/63941, kern/66124 Submitted by: Magnus <bsdhead.at.spray.dot.se@FreeBSD.org>, Dmitry Dyomin <old@old.com.ua>, Dmitry Sivachenko <mitya@demos.su> MFC after: 1 week
2004-08-08 09:08:37 +00:00
{
/*
* iRiver iFP MP3 player (with UMS Firmware)
* PR: kern/54881, i386/63941, kern/66124
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "iRiver", "iFP*", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE
add support Frontier Labs NEX IA+ Digital Audio Player with USB CF card reader/writer PR: kern/70158 Submitted by: Bernd Strau. <no_bs@web.de> MFC after: 1 week
2004-08-12 23:17:09 +00:00
},
{
/*
* Frontier Labs NEX IA+ Digital Audio Player, rev 1.10/0.01
* PR: kern/70158
*/
Apply quirk suggested by submitter. PR: 75486 Submitted by: no_bs@web.de MFC after: 1 week
2005-04-14 03:59:48 +00:00
{T_DIRECT, SIP_MEDIA_REMOVABLE, "FL" , "Nex*", "*"},
add support Frontier Labs NEX IA+ Digital Audio Player with USB CF card reader/writer PR: kern/70158 Submitted by: Bernd Strau. <no_bs@web.de> MFC after: 1 week
2004-08-12 23:17:09 +00:00
/*quirks*/ DA_Q_NO_SYNC_CACHE
},
Quirk for ZICPlay USB MP3 Player. PR: kern/75057 Submitted by: Aurelien Nephtali <aurelien.nephtali wanadoo.fr>
2005-04-14 04:31:48 +00:00
{
/*
* ZICPlay USB MP3 Player with FM
* PR: kern/75057
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "ACTIONS*" , "USB DISK*", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE
},
Add quirk for TEAC USB floppy drives.
2005-05-05 18:48:41 +00:00
{
/*
* TEAC USB floppy mechanisms
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "TEAC" , "FD-05*", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE
},
Add a quirk for my pen-drive.
2005-06-09 17:35:04 +00:00
{
/*
* Kingston DataTraveler II+ USB Pen-Drive.
* Reported by: Pawel Jakub Dawidek <pjd@FreeBSD.org>
*/
Style nit.
2005-12-19 03:43:48 +00:00
{T_DIRECT, SIP_MEDIA_REMOVABLE, "Kingston" , "DataTraveler II+",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
Add a quirk for my pen-drive.
2005-06-09 17:35:04 +00:00
},
Add support for the USB DISK Pro PMAP. This patch has sit for 6 years in the PR database. PR: usb/96381 Submitted by: jhs Reviewed by: mav Approved by: cperciva (implicit) MFC after: 3 days
2012-10-15 12:33:53 +00:00
{
/*
* USB DISK Pro PMAP
* Reported by: jhs
* PR: usb/96381
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, " ", "USB DISK Pro", "PMAP"},
/*quirks*/ DA_Q_NO_SYNC_CACHE
},
Allow to use TransFlash drive, which can be found in Motorola E398 Mobile Phone. PR: usb/89889 Submitted by: Wojciech A. Koszek <dunstan@freebsd.czest.pl> MFC after: 1 week
2005-12-19 03:27:28 +00:00
{
/*
* Motorola E398 Mobile Phone (TransFlash memory card).
* Reported by: Wojciech A. Koszek <dunstan@FreeBSD.czest.pl>
* PR: usb/89889
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "Motorola" , "Motorola Phone",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
Allow to use Time DPA20B 1GB MP3 Player [1] Allow to use Qware BeatZkey! Pro [2] PR: usb/81846 [1], usb/79164 [2] Submitted by: Ruben de Groot <rdg@bzerk.org> [1], Radek Kozlowski <radek@raadradd.com> [2] Approved by: njl MFC: 1 week
2006-01-13 11:33:40 +00:00
{
/*
* Qware BeatZkey! Pro
* PR: usb/79164
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "GENERIC", "USB DISK DEVICE",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
{
/*
* Time DPA20B 1GB MP3 Player
* PR: usb/81846
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "USB2.0*", "(FS) FLASH DISK*",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
Add Product IDs : - Sandisk Cruzer Micro 128MB [5] - DANE-ELEC zMate 512MB USB flash drive [7] - Attache 256MB USB 2.0 Flash Drive [8] - Sandisk Cruzer Micro 256MB [9] Add scsi_da.c quirks : - Samsung USB key 128Mb [1] - Kingston DataTraveler 2.0 [2] - Creative MuVo Slim [3] - United MP 5512 Portable MP3 Player [4] - Sandisk Cruzer Micro 128MB [5], [9] - PNY USB Flash keys [6], [7], [8] Add umass.c quirks : - Sandisk Cruzer Micro 128MB [5] - DANE-ELEC zMate 512MB USB flash drive [7] - Attache 256MB USB 2.0 Flash Drive [8] - Sandisk Cruzer Micro 256MB [9] PR: usb/90081 [1], usb/89196 [2], kern/86131 [3], usb/80487 [4], usb/75970 [5], usb/75578 [6], usb/72344 [7], usb/65436 [8], usb/70835 [9] Submitted by: Henri-Pierre CHARLES <hpc@prism.uvsq.fr> [1], Andrey Simonenko <simon@comsys.ntu-kpi.kiev.ua> [2], Erich Rickheit KSC <rickheit-fbp@numachi.com> [3], tnu@chania.di.uoa.gr [4], Bram Abbekerk <bram@abbekerk.demon.nl> [5], Thomas Pornin <pornin@bolet.org> [6], parv <parv@pair.com> [7], Peter D. Quilty <pdquilty@adelphia.net> [8], Raymundo M. Vega <rvega@ms-smtp-03-eri0.socal.rr.com> [9] Approved by: iedowse MFC after: 3 days
2006-01-26 00:35:54 +00:00
{
/*
* Samsung USB key 128Mb
* PR: usb/90081
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "USB-DISK", "FreeDik-FlashUsb",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
{
/*
* Kingston DataTraveler 2.0 USB Flash memory.
* PR: usb/89196
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "Kingston", "DataTraveler 2.0",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
{
/*
* Creative MUVO Slim mp3 player (USB)
* PR: usb/86131
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "CREATIVE", "MuVo Slim",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE|DA_Q_NO_PREVENT
},
{
/*
* United MP5512 Portable MP3 Player (2-in-1 USB DISK/MP3)
* PR: usb/80487
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "Generic*", "MUSIC DISK",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
{
/*
* SanDisk Micro Cruzer 128MB
* PR: usb/75970
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "SanDisk" , "Micro Cruzer",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
Add the TOSHIBA TransMemory USB sticks to the list of devices that hate "Synchronize cache" commands. MFC after: 1 day
2006-03-13 22:26:33 +00:00
{
/*
* TOSHIBA TransMemory USB sticks
Add reference to PR to TOSHIBA TransMemory quirk entry.
2006-03-18 21:13:14 +00:00
* PR: kern/94660
Add the TOSHIBA TransMemory USB sticks to the list of devices that hate "Synchronize cache" commands. MFC after: 1 day
2006-03-13 22:26:33 +00:00
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "TOSHIBA", "TransMemory",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
Add quirks to disable READ CAPACITY (16) for PNY USB 3.0 Flash Drives. Submitted by: Sean Fagan <sef@ixsystems.com> MFC after: 3 days
2014-09-15 19:48:27 +00:00
{
/*
* PNY USB 3.0 Flash Drives
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "PNY", "USB 3.0 FD*",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE | DA_Q_NO_RC16
},
Add Product IDs : - Sandisk Cruzer Micro 128MB [5] - DANE-ELEC zMate 512MB USB flash drive [7] - Attache 256MB USB 2.0 Flash Drive [8] - Sandisk Cruzer Micro 256MB [9] Add scsi_da.c quirks : - Samsung USB key 128Mb [1] - Kingston DataTraveler 2.0 [2] - Creative MuVo Slim [3] - United MP 5512 Portable MP3 Player [4] - Sandisk Cruzer Micro 128MB [5], [9] - PNY USB Flash keys [6], [7], [8] Add umass.c quirks : - Sandisk Cruzer Micro 128MB [5] - DANE-ELEC zMate 512MB USB flash drive [7] - Attache 256MB USB 2.0 Flash Drive [8] - Sandisk Cruzer Micro 256MB [9] PR: usb/90081 [1], usb/89196 [2], kern/86131 [3], usb/80487 [4], usb/75970 [5], usb/75578 [6], usb/72344 [7], usb/65436 [8], usb/70835 [9] Submitted by: Henri-Pierre CHARLES <hpc@prism.uvsq.fr> [1], Andrey Simonenko <simon@comsys.ntu-kpi.kiev.ua> [2], Erich Rickheit KSC <rickheit-fbp@numachi.com> [3], tnu@chania.di.uoa.gr [4], Bram Abbekerk <bram@abbekerk.demon.nl> [5], Thomas Pornin <pornin@bolet.org> [6], parv <parv@pair.com> [7], Peter D. Quilty <pdquilty@adelphia.net> [8], Raymundo M. Vega <rvega@ms-smtp-03-eri0.socal.rr.com> [9] Approved by: iedowse MFC after: 3 days
2006-01-26 00:35:54 +00:00
{
/*
* PNY USB Flash keys
* PR: usb/75578, usb/72344, usb/65436
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "*" , "USB DISK*",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
- Add a scsi_da.c and a umass.c quirk for Genesys 6-in-1 Card Reader. Reported by: anders on freebsd-usb@ Tested by: anders Approved by: ssouhlal MFC after: 3 days
2006-01-30 20:27:44 +00:00
{
/*
* Genesys 6-in-1 Card Reader
- Add a PR number for future reference. Approved by: njl
2006-03-18 18:28:13 +00:00
* PR: usb/94647
- Add a scsi_da.c and a umass.c quirk for Genesys 6-in-1 Card Reader. Reported by: anders on freebsd-usb@ Tested by: anders Approved by: ssouhlal MFC after: 3 days
2006-01-30 20:27:44 +00:00
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "Generic*", "STORAGE DEVICE*",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
Add a SCSI da(4) quirk for Rekam USB Camera. PR: usb/98713 Submitted by: Alex <goo@t72.ru> MFC after: 3 days
2006-10-07 19:04:23 +00:00
{
/*
* Rekam Digital CAMERA
* PR: usb/98713
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "CAMERA*", "4MP-9J6*",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
Add SCSI da(4) quirk for the iRiver H10. PR: usb/102547 Submitted by: Sven Esbjerg <freebsd-pr@xbsd.net> MFC after: 3 days
2006-10-07 18:28:55 +00:00
{
/*
* iRiver H10 MP3 player
* PR: usb/102547
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "iriver", "H10*",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
o Style(9): trim eol w/spaces, w/spaces to tabs, kill an empty line.
2006-11-26 17:15:54 +00:00
},
Support for iRiver U10 USB connection PR: 92306 Submitted by: Soren Dossing Approved by: re (blanket)
2007-06-28 06:12:55 +00:00
{
/*
* iRiver U10 MP3 player
* PR: usb/92306
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "iriver", "U10*",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
o Add a quirk for X-Micro Flash Disk. PR: usb/96901 Submitted by: Szalai Andras Reviewed by: nate (a half of year ago) MFC after: 3 weeks
2006-11-26 17:13:24 +00:00
{
/*
* X-Micro Flash Disk
* PR: usb/96901
*/
o Style: remove a w/space before a comma.
2006-11-26 17:18:31 +00:00
{T_DIRECT, SIP_MEDIA_REMOVABLE, "X-Micro", "Flash Disk",
o Add a quirk for X-Micro Flash Disk. PR: usb/96901 Submitted by: Szalai Andras Reviewed by: nate (a half of year ago) MFC after: 3 weeks
2006-11-26 17:13:24 +00:00
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
Add quirk for EasyMP3 EM732X usb 2.0 flash mp3 player. (It appears that the quirk proceedures link has disappeared and that this PR complied with it, if there's a problem, please contact me). PR: usb/96546
2007-01-22 04:34:03 +00:00
{
/*
* EasyMP3 EM732X USB 2.0 Flash MP3 Player
* PR: usb/96546
*/
Use the more specific 'EM732X' designation rather than * to disable sync cache commands, per request from njl@.
2007-01-23 17:29:31 +00:00
{T_DIRECT, SIP_MEDIA_REMOVABLE, "EM732X", "MP3 Player*",
o Correct EasyMP3 EM732X usb 2.0 flash mp3 player revision: it's 1.00 not 1.0. PR: usb/96546 Submitted by: Eugene Grosbein MFC after: 1 month
2008-01-12 19:24:55 +00:00
"1.00"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
Add quirk for EasyMP3 EM732X usb 2.0 flash mp3 player. (It appears that the quirk proceedures link has disappeared and that this PR complied with it, if there's a problem, please contact me). PR: usb/96546
2007-01-22 04:34:03 +00:00
},
Quirk for Denver MP3 player usb. Submitted by: Ed Schouten PR: 107101 Approved by: re (blanket for device supprot)
2007-06-28 04:51:19 +00:00
{
/*
* Denver MP3 player
* PR: usb/107101
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "DENVER", "MP3 PLAYER",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
Philips USB Key Audio KEY013 Submitted by: Arnoud Engelfriet PR: 68412 Approved by: re (blanket)
2007-06-30 04:34:45 +00:00
{
/*
* Philips USB Key Audio KEY013
* PR: usb/68412
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "PHILIPS", "Key*", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE | DA_Q_NO_PREVENT
},
Add support for JNC MP3 Player Submitted by: User & <dhenin@cypries.cyclopes.org> PR: 94439 Approved by: re (blanket)
2007-06-30 05:03:03 +00:00
{
/*
* JNC MP3 Player
* PR: usb/94439
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "JNC*" , "MP3 Player*",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
Samsung MP0402H quirk Submitted by: Gipsy<wint3r@securimail.com> PR: 108427 Approved by: re@ (blanket)
2007-06-30 05:21:18 +00:00
{
/*
* SAMSUNG MP0402H
* PR: usb/108427
*/
{T_DIRECT, SIP_MEDIA_FIXED, "SAMSUNG", "MP0402H", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE
},
quirk for I/O Magic USB flash drive "Giga Bank" Submitted by: Jeff Anton PR: 108810 Approved by: re@ (blanket)
2007-06-30 05:24:25 +00:00
{
/*
* I/O Magic USB flash - Giga Bank
* PR: usb/108810
*/
{T_DIRECT, SIP_MEDIA_FIXED, "GS-Magic", "stor*", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE
},
Add quirk for JoyFly 128mb USB Flash Drive submitted by: Toni Schmidbauer PR: 96133 Approved by: re@ (blanket)
2007-06-30 14:41:30 +00:00
{
/*
* JoyFly 128mb USB Flash Drive
* PR: 96133
*/
After talking with njl@, change this wildcard entry to be a bit less wild. Approved by: re (blanket)
2007-07-01 17:42:07 +00:00
{T_DIRECT, SIP_MEDIA_REMOVABLE, "USB 2.0", "Flash Disk*",
Add quirk for JoyFly 128mb USB Flash Drive submitted by: Toni Schmidbauer PR: 96133 Approved by: re@ (blanket)
2007-06-30 14:41:30 +00:00
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
Quirk for ChipsBnk usb stick Submitted by: Mark Andrews PR: 103702 Approved by: re@ (blanket)
2007-06-30 14:44:09 +00:00
{
/*
* ChipsBnk usb stick
* PR: 103702
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "ChipsBnk", "USB*",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
Add quirk for the Storcase InfoStation 12bay SATA to FC SAN. PR: 129858 Submitted by: Nick Triantos <nick-freebsd at triantos dot com> MFC after: 1 week
2008-12-23 09:11:05 +00:00
{
/*
* Storcase (Kingston) InfoStation IFS FC2/SATA-R 201A
* PR: 129858
*/
{T_DIRECT, SIP_MEDIA_FIXED, "IFS", "FC2/SATA-R*",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
Add quirk for the Samsung YP-U3 PR: 125398 Submitted by: Tino Engel <goaengel at gmx dot net> Approved by: imp (mentor, implicit) MFC after: 1 week
2008-07-09 17:00:14 +00:00
{
/*
* Samsung YP-U3 mp3-player
* PR: 125398
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "Samsung", "YP-U3",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
Add quirks for Netac Onlydisk 2000 USB disk.
2007-12-11 06:41:38 +00:00
{
{T_DIRECT, SIP_MEDIA_REMOVABLE, "Netac", "OnlyDisk*",
"2000"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
- Add quirk for Sony DSC digital cameras. This umass devices fail to attach without these quirks applied. PR: usb/137035 URL: http://lists.freebsd.org/pipermail/freebsd-current/2009-August/010852.html Reported by: Henri Hennebert <hlh@restart.be>, Andrey V. Elsukov <bu7cher@yandex.ru> MFC after: 1 week
2009-08-26 21:14:28 +00:00
},
{
/*
* Sony Cyber-Shot DSC cameras
* PR: usb/137035
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "Sony", "Sony DSC", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE | DA_Q_NO_PREVENT
Merge to da driver quirks hinting 4K physical sector sizes for SATA disks connected via SAS or USB. Unluckily I've found that SAS (mps) and USB-SATA I have translate models in different ways, requiring twice more quirks. Unluckily for Hitachi, their model names are trimmed on SAS, making impossible to identify 4K sector drives that way.
2011-12-23 00:10:17 +00:00
},
Add DA_Q_NO_PREVENT quirk for Kingston DataTraveler G3 1.00 USB flash. PREVENT ALLOW MEDIUM REMOVAL commands return errors on these devices without returning sense data. In some cases unrelated following commands start to return errors too, that makes device to be dropped by CAM.
2013-02-22 17:45:32 +00:00
{
{T_DIRECT, SIP_MEDIA_REMOVABLE, "Kingston", "DataTraveler G3",
"1.00"}, /*quirks*/ DA_Q_NO_PREVENT
},
Add NO_RC16 quirk to make da driver avoid using READ CAPACITY(16) command if possible. Use it for Kingston JetFlash USB sticks, that are known to return garbage in response to that command.
2013-07-30 13:00:09 +00:00
{
/* At least several Transcent USB sticks lie on RC16. */
{T_DIRECT, SIP_MEDIA_REMOVABLE, "JetFlash", "Transcend*",
"*"}, /*quirks*/ DA_Q_NO_RC16
},
Merge to da driver quirks hinting 4K physical sector sizes for SATA disks connected via SAS or USB. Unluckily I've found that SAS (mps) and USB-SATA I have translate models in different ways, requiring twice more quirks. Unluckily for Hitachi, their model names are trimmed on SAS, making impossible to identify 4K sector drives that way.
2011-12-23 00:10:17 +00:00
/* ATA/SATA devices over SAS/USB/... */
{
/* Hitachi Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "Hitachi", "H??????????E3*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Samsung Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "SAMSUNG HD155UI*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Samsung Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "SAMSUNG", "HD155UI*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Samsung Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "SAMSUNG HD204UI*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Samsung Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "SAMSUNG", "HD204UI*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Barracuda Green Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "ST????DL*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Barracuda Green Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ST????DL", "*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Barracuda Green Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "ST???DM*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Barracuda Green Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ST???DM*", "*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Barracuda Green Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "ST????DM*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Barracuda Green Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ST????DM", "*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Momentus Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "ST9500423AS*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Momentus Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ST950042", "3AS*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Momentus Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "ST9500424AS*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Momentus Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ST950042", "4AS*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Momentus Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "ST9640423AS*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Momentus Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ST964042", "3AS*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Momentus Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "ST9640424AS*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Momentus Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ST964042", "4AS*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Momentus Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "ST9750420AS*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Momentus Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ST975042", "0AS*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Momentus Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "ST9750422AS*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Momentus Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ST975042", "2AS*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Momentus Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "ST9750423AS*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Momentus Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ST975042", "3AS*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Momentus Thin Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "ST???LT*", "*" },
/*quirks*/DA_Q_4K
},
{
/* Seagate Momentus Thin Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ST???LT*", "*", "*" },
/*quirks*/DA_Q_4K
},
{
/* WDC Caviar Green Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "WDC WD????RS*", "*" },
/*quirks*/DA_Q_4K
},
{
/* WDC Caviar Green Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "WDC WD??", "??RS*", "*" },
/*quirks*/DA_Q_4K
},
{
/* WDC Caviar Green Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "WDC WD????RX*", "*" },
/*quirks*/DA_Q_4K
},
{
/* WDC Caviar Green Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "WDC WD??", "??RX*", "*" },
/*quirks*/DA_Q_4K
},
{
/* WDC Caviar Green Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "WDC WD??????RS*", "*" },
/*quirks*/DA_Q_4K
},
{
/* WDC Caviar Green Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "WDC WD??", "????RS*", "*" },
/*quirks*/DA_Q_4K
},
{
/* WDC Caviar Green Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "WDC WD??????RX*", "*" },
/*quirks*/DA_Q_4K
},
{
/* WDC Caviar Green Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "WDC WD??", "????RX*", "*" },
/*quirks*/DA_Q_4K
},
{
/* WDC Scorpio Black Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "WDC WD???PKT*", "*" },
/*quirks*/DA_Q_4K
},
{
/* WDC Scorpio Black Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "WDC WD??", "?PKT*", "*" },
/*quirks*/DA_Q_4K
},
{
/* WDC Scorpio Black Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "WDC WD?????PKT*", "*" },
/*quirks*/DA_Q_4K
},
{
/* WDC Scorpio Black Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "WDC WD??", "???PKT*", "*" },
/*quirks*/DA_Q_4K
},
{
/* WDC Scorpio Blue Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "WDC WD???PVT*", "*" },
/*quirks*/DA_Q_4K
},
{
/* WDC Scorpio Blue Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "WDC WD??", "?PVT*", "*" },
/*quirks*/DA_Q_4K
},
{
/* WDC Scorpio Blue Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "WDC WD?????PVT*", "*" },
/*quirks*/DA_Q_4K
},
{
/* WDC Scorpio Blue Advanced Format (4k) drives */
{ T_DIRECT, SIP_MEDIA_FIXED, "WDC WD??", "???PVT*", "*" },
/*quirks*/DA_Q_4K
},
Add support for: Olympus FE-210 camera LG UP3S MP3 player Laser MP3-2GA13 MP3 PR: usb/119201 Submitted by: Peter Jeremy <peterjeremy@optushome.com.au> Approved by: cperciva MFC after: 1 week
2012-04-29 22:28:15 +00:00
{
/*
* Olympus FE-210 camera
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "OLYMPUS", "FE210*",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
{
/*
* LG UP3S MP3 player
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "LG", "UP3S",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
{
/*
* Laser MP3-2GA13 MP3 player
*/
{T_DIRECT, SIP_MEDIA_REMOVABLE, "USB 2.0", "(HS) Flash Disk",
"*"}, /*quirks*/ DA_Q_NO_SYNC_CACHE
},
Add support for samsung HM250JI PR: usb/121474 Submitted by: Ben Stuyts <ben@altesco.nl> Approved by: cperciva (implicit) MFC after: 3 days
2012-10-16 17:49:14 +00:00
{
/*
* LaCie external 250GB Hard drive des by Porsche
* Submitted by: Ben Stuyts <ben@altesco.nl>
* PR: 121474
*/
{T_DIRECT, SIP_MEDIA_FIXED, "SAMSUNG", "HM250JI", "*"},
/*quirks*/ DA_Q_NO_SYNC_CACHE
},
Added missing SCSI quirks from r241784 Re-ordered SSD quirks alphabetically so they are easier to maintain. Removed my email and PR reference from comments on each quirk. Added quirks for more SSDs: * Crucial M4 * Corsair Force GT * Intel 520 Series * Kingston E100 Series * Samsung 830 Series Reviewed by: pjd (mentor) Approved by: pjd (mentor) MFC after: 1 week
2013-05-28 14:44:37 +00:00
/* SATA SSDs */
{
/*
* Corsair Force 2 SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "Corsair CSSD-F*", "*" },
/*quirks*/DA_Q_4K
},
{
/*
* Corsair Force 3 SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "Corsair Force 3*", "*" },
/*quirks*/DA_Q_4K
},
Added 4K quirks for Corsair Neutron GTX SSD's
2013-10-15 17:03:02 +00:00
{
/*
* Corsair Neutron GTX SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "*", "Corsair Neutron GTX*", "*" },
/*quirks*/DA_Q_4K
},
Added missing SCSI quirks from r241784 Re-ordered SSD quirks alphabetically so they are easier to maintain. Removed my email and PR reference from comments on each quirk. Added quirks for more SSDs: * Crucial M4 * Corsair Force GT * Intel 520 Series * Kingston E100 Series * Samsung 830 Series Reviewed by: pjd (mentor) Approved by: pjd (mentor) MFC after: 1 week
2013-05-28 14:44:37 +00:00
{
/*
Added 4K quirks for Corsair Force GT and Samsung 840 SSDs MFC after: 1 week Sponsored by: Multiplay
2014-08-14 13:57:17 +00:00
* Corsair Force GT & GS SSDs
Added missing SCSI quirks from r241784 Re-ordered SSD quirks alphabetically so they are easier to maintain. Removed my email and PR reference from comments on each quirk. Added quirks for more SSDs: * Crucial M4 * Corsair Force GT * Intel 520 Series * Kingston E100 Series * Samsung 830 Series Reviewed by: pjd (mentor) Approved by: pjd (mentor) MFC after: 1 week
2013-05-28 14:44:37 +00:00
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
Added 4K quirks for Corsair Force GT and Samsung 840 SSDs MFC after: 1 week Sponsored by: Multiplay
2014-08-14 13:57:17 +00:00
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "Corsair Force G*", "*" },
Added missing SCSI quirks from r241784 Re-ordered SSD quirks alphabetically so they are easier to maintain. Removed my email and PR reference from comments on each quirk. Added quirks for more SSDs: * Crucial M4 * Corsair Force GT * Intel 520 Series * Kingston E100 Series * Samsung 830 Series Reviewed by: pjd (mentor) Approved by: pjd (mentor) MFC after: 1 week
2013-05-28 14:44:37 +00:00
/*quirks*/DA_Q_4K
},
{
/*
* Crucial M4 SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "M4-CT???M4SSD2*", "*" },
/*quirks*/DA_Q_4K
},
{
/*
* Crucial RealSSD C300 SSDs
* 4k optimised
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "C300-CTFDDAC???MAG*",
"*" }, /*quirks*/DA_Q_4K
},
{
/*
* Intel 320 Series SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "INTEL SSDSA2CW*", "*" },
/*quirks*/DA_Q_4K
},
{
/*
* Intel 330 Series SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "INTEL SSDSC2CT*", "*" },
/*quirks*/DA_Q_4K
},
{
/*
* Intel 510 Series SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "INTEL SSDSC2MH*", "*" },
/*quirks*/DA_Q_4K
},
{
/*
* Intel 520 Series SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "INTEL SSDSC2BW*", "*" },
/*quirks*/DA_Q_4K
},
Added 4K quirks for:- * OCZ Agility 2 SSDs * Marvell SSDs * Intel X25-M Series SSDs
2013-08-14 15:18:28 +00:00
{
/*
* Intel X25-M Series SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "INTEL SSDSA2M*", "*" },
/*quirks*/DA_Q_4K
},
Added missing SCSI quirks from r241784 Re-ordered SSD quirks alphabetically so they are easier to maintain. Removed my email and PR reference from comments on each quirk. Added quirks for more SSDs: * Crucial M4 * Corsair Force GT * Intel 520 Series * Kingston E100 Series * Samsung 830 Series Reviewed by: pjd (mentor) Approved by: pjd (mentor) MFC after: 1 week
2013-05-28 14:44:37 +00:00
{
/*
* Kingston E100 Series SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "KINGSTON SE100S3*", "*" },
/*quirks*/DA_Q_4K
},
{
/*
* Kingston HyperX 3k SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "KINGSTON SH103S3*", "*" },
/*quirks*/DA_Q_4K
},
Added 4K quirks for:- * OCZ Agility 2 SSDs * Marvell SSDs * Intel X25-M Series SSDs
2013-08-14 15:18:28 +00:00
{
/*
* Marvell SSDs (entry taken from OpenSolaris)
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "MARVELL SD88SA02*", "*" },
/*quirks*/DA_Q_4K
},
{
/*
* OCZ Agility 2 SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "*", "OCZ-AGILITY2*", "*" },
/*quirks*/DA_Q_4K
},
Added missing SCSI quirks from r241784 Re-ordered SSD quirks alphabetically so they are easier to maintain. Removed my email and PR reference from comments on each quirk. Added quirks for more SSDs: * Crucial M4 * Corsair Force GT * Intel 520 Series * Kingston E100 Series * Samsung 830 Series Reviewed by: pjd (mentor) Approved by: pjd (mentor) MFC after: 1 week
2013-05-28 14:44:37 +00:00
{
/*
* OCZ Agility 3 SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "OCZ-AGILITY3*", "*" },
/*quirks*/DA_Q_4K
},
{
/*
* OCZ Deneva R Series SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "DENRSTE251M45*", "*" },
/*quirks*/DA_Q_4K
},
{
/*
* OCZ Vertex 2 SSDs (inc pro series)
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "OCZ?VERTEX2*", "*" },
/*quirks*/DA_Q_4K
},
{
/*
* OCZ Vertex 3 SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "OCZ-VERTEX3*", "*" },
/*quirks*/DA_Q_4K
},
Added 4K QUIRK for OCZ Vertex 4 SSDs Submitted by: Borja Marcos <borjam@sarenet.es> MFC after: 2 days
2013-07-09 10:41:17 +00:00
{
/*
* OCZ Vertex 4 SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "OCZ-VERTEX4*", "*" },
/*quirks*/DA_Q_4K
},
Added missing SCSI quirks from r241784 Re-ordered SSD quirks alphabetically so they are easier to maintain. Removed my email and PR reference from comments on each quirk. Added quirks for more SSDs: * Crucial M4 * Corsair Force GT * Intel 520 Series * Kingston E100 Series * Samsung 830 Series Reviewed by: pjd (mentor) Approved by: pjd (mentor) MFC after: 1 week
2013-05-28 14:44:37 +00:00
{
/*
* Samsung 830 Series SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "SAMSUNG SSD 830 Series*", "*" },
/*quirks*/DA_Q_4K
},
Added 4K quirks for Corsair Force GT and Samsung 840 SSDs MFC after: 1 week Sponsored by: Multiplay
2014-08-14 13:57:17 +00:00
{
/*
* Samsung 840 SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "Samsung SSD 840*", "*" },
/*quirks*/DA_Q_4K
},
Add the Samsung 843T as a 4k enabled drive Submitted by: Jason Wolfe <jason@llnw.com> MFC after: 2 weeks Sponsored by: Limelight Networks
2014-08-21 21:05:58 +00:00
{
/*
* Samsung 843T Series SSDs
* 4k optimised
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "SAMSUNG MZ7WD*", "*" },
/*quirks*/DA_Q_4K
},
Add new quirks for the latest Samsung SSD, model 850. Submitted by: sbruno MFC after: 2 weeks
2014-10-19 16:46:36 +00:00
{
/*
* Samsung 850 SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "Samsung SSD 850*", "*" },
/*quirks*/DA_Q_4K
},
Add 4k quirks for PM853T Samsung SSD MFC after: 2 weeks Sponsored by: Limelight Networks
2014-10-16 20:33:04 +00:00
{
/*
* Samsung PM853T Series SSDs
* 4k optimised
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "SAMSUNG MZ7GE*", "*" },
/*quirks*/DA_Q_4K
},
Add 4k enabled cam quirks for Samsung SM863 Series SSDs Submitted by: Jason (j@nitrology.com) MFC after: 2 weeks Sponsored by: Limelight Networks Differential Revision: https://reviews.freebsd.org/D5711
2016-03-24 14:20:33 +00:00
{
/*
* Samsung SM863 Series SSDs
* 4k optimised
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "SAMSUNG MZ7KM*", "*" },
/*quirks*/DA_Q_4K
},
Added missing SCSI quirks from r241784 Re-ordered SSD quirks alphabetically so they are easier to maintain. Removed my email and PR reference from comments on each quirk. Added quirks for more SSDs: * Crucial M4 * Corsair Force GT * Intel 520 Series * Kingston E100 Series * Samsung 830 Series Reviewed by: pjd (mentor) Approved by: pjd (mentor) MFC after: 1 week
2013-05-28 14:44:37 +00:00
{
/*
* SuperTalent TeraDrive CT SSDs
* 4k optimised & trim only works in 4k requests + 4k aligned
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "FTM??CT25H*", "*" },
/*quirks*/DA_Q_4K
},
{
/*
* XceedIOPS SATA SSDs
* 4k optimised
*/
{ T_DIRECT, SIP_MEDIA_FIXED, "ATA", "SG9XCS2D*", "*" },
/*quirks*/DA_Q_4K
},
Add DA_Q_NO_RC16 quirk for USB mass storage device. PR: 194062 MFC after: 1 week
2015-03-07 17:18:06 +00:00
{
/*
* Hama Innostor USB-Stick
*/
{ T_DIRECT, SIP_MEDIA_REMOVABLE, "Innostor", "Innostor*", "*" },
/*quirks*/DA_Q_NO_RC16
},
Add DA_Q_NO_RC16 quirk for USB mass storage device. PR: 198647 MFC after: 1 week
2015-03-25 13:28:13 +00:00
{
/*
* MX-ES USB Drive by Mach Xtreme
*/
Extend DA_Q_NO_RC16 to MXUB3* devices. PR: kern/198647 MFC after: 2 weeks
2015-04-21 22:55:52 +00:00
{ T_DIRECT, SIP_MEDIA_REMOVABLE, "MX", "MXUB3*", "*"},
Add DA_Q_NO_RC16 quirk for USB mass storage device. PR: 198647 MFC after: 1 week
2015-03-25 13:28:13 +00:00
/*quirks*/DA_Q_NO_RC16
},
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
};
NO_GEOM cleanup: Move to "struct disk *" centric api.
2003-02-25 22:06:21 +00:00
static disk_strategy_t dastrategy;
NO_GEOM cleanup: Retire the "d_dump_t" and use the "dumper_t" type instead. Dumper_t takes a void * as first arg which is more general than the dev_t taken by d_dump_t. (Remember: we could have net-dumpers if somebody wrote us one!) Define the convention for GEOM controlled disk devices to be that the first argument to the dumper function is the struct disk pointer. Change device drivers accordingly.
2003-02-21 19:00:48 +00:00
static dumper_t dadump;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
static periph_init_t dainit;
static void daasync(void *callback_arg, u_int32_t code,
struct cam_path *path, void *arg);
Move dynamic sysctl(8) variable creation for the cd(4) and da(4) drivers out of cdregister() and daregister(), which are run from interrupt context. The sysctl code does blocking mallocs (M_WAITOK), which causes problems if malloc(9) actually needs to sleep. The eventual fix for this issue will involve moving the CAM probe process inside a kernel thread. For now, though, I have fixed the issue by moving dynamic sysctl variable creation for these two drivers to a task queue running in a kernel thread. The existing task queues (taskqueue_swi and taskqueue_swi_giant) run in software interrupt handlers, which wouldn't fix the problem at hand. So I have created a new task queue, taskqueue_thread, that runs inside a kernel thread. (It also runs outside of Giant -- clients must explicitly acquire and release Giant in their taskqueue functions.) scsi_cd.c: Remove sysctl variable creation code from cdregister(), and move it to a new function, cdsysctlinit(). Queue cdsysctlinit() to the taskqueue_thread taskqueue once we have fully registered the cd(4) driver instance. scsi_da.c: Remove sysctl variable creation code from daregister(), and move it to move it to a new function, dasysctlinit(). Queue dasysctlinit() to the taskqueue_thread taskqueue once we have fully registered the da(4) instance. taskqueue.h: Declare the new taskqueue_thread taskqueue, update some comments. subr_taskqueue.c: Create the new kernel thread taskqueue. This taskqueue runs outside of Giant, so any functions queued to it would need to explicitly acquire/release Giant if they need it. cd.4: Update the cd(4) man page to talk about the minimum command size sysctl/loader tunable. Also note that the changer variables are available as loader tunables as well. da.4: Update the da(4) man page to cover the retry_count, default_timeout and minimum_cmd_size sysctl variables/loader tunables. Remove references to /dev/r???, they aren't used any longer. cd.9: Update the cd(9) man page to describe the CD_Q_10_BYTE_ONLY quirk. taskqueue.9: Update the taskqueue(9) man page to describe the new thread task queue, and the taskqueue_swi_giant queue. MFC after: 3 days
2003-09-03 04:46:28 +00:00
static void dasysctlinit(void *context, int pending);
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
static int dacmdsizesysctl(SYSCTL_HANDLER_ARGS);
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
static int dadeletemethodsysctl(SYSCTL_HANDLER_ARGS);
Teach GEOM and CAM about the difference between the max "size" of r/w and delete requests. sys/geom/geom_disk.h: - Added d_delmaxsize which represents the maximum size of individual device delete requests in bytes. This can be used by devices to inform geom of their size limitations regarding delete operations which are generally different from the read / write limits as data is not usually transferred from the host to physical device. sys/geom/geom_disk.c: - Use new d_delmaxsize to calculate the size of chunks passed through to the underlying strategy during deletes instead of using read / write optimised values. This defaults to d_maxsize if unset (0). - Moved d_maxsize default up so it can be used to default d_delmaxsize sys/cam/ata/ata_da.c: - Added d_delmaxsize calculations for TRIM and CFA sys/cam/scsi/scsi_da.c: - Added re-calculation of d_delmaxsize whenever delete_method is set. - Added kern.cam.da.X.delete_max sysctl which allows the max size for delete requests to be limited. This is useful in preventing timeouts on devices who's delete methods are slow. It should be noted that this limit is reset then the device delete method is changed and that it can only be lowered not increased from the device max. Reviewed by: mav Approved by: pjd (mentor)
2013-04-26 16:22:54 +00:00
static int dadeletemaxsysctl(SYSCTL_HANDLER_ARGS);
Removed unneeded tests in dadeletemethodset changing it to return void Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 15:31:52 +00:00
static void dadeletemethodset(struct da_softc *softc,
Updates delete_method sysctl changes to always maintain disk d_flags DISKFLAG_CANDELETE. While this change makes this layer consistent other layers such as UFS and ZFS BIO_DELETE support may not notice any change made manually via these device sysctls until the device is reopened via a mount. Also corrected var order in dadeletemethodsysctl PR: kern/169801 Reviewed by: pjd (mentor) Approved by: mav MFC after: 2 weeks
2013-01-10 11:57:46 +00:00
da_delete_methods delete_method);
Teach GEOM and CAM about the difference between the max "size" of r/w and delete requests. sys/geom/geom_disk.h: - Added d_delmaxsize which represents the maximum size of individual device delete requests in bytes. This can be used by devices to inform geom of their size limitations regarding delete operations which are generally different from the read / write limits as data is not usually transferred from the host to physical device. sys/geom/geom_disk.c: - Use new d_delmaxsize to calculate the size of chunks passed through to the underlying strategy during deletes instead of using read / write optimised values. This defaults to d_maxsize if unset (0). - Moved d_maxsize default up so it can be used to default d_delmaxsize sys/cam/ata/ata_da.c: - Added d_delmaxsize calculations for TRIM and CFA sys/cam/scsi/scsi_da.c: - Added re-calculation of d_delmaxsize whenever delete_method is set. - Added kern.cam.da.X.delete_max sysctl which allows the max size for delete requests to be limited. This is useful in preventing timeouts on devices who's delete methods are slow. It should be noted that this limit is reset then the device delete method is changed and that it can only be lowered not increased from the device max. Reviewed by: mav Approved by: pjd (mentor)
2013-04-26 16:22:54 +00:00
static off_t dadeletemaxsize(struct da_softc *softc,
da_delete_methods delete_method);
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
static void dadeletemethodchoose(struct da_softc *softc,
da_delete_methods default_method);
Use the existence of ATA Information VPD to determine if we should attempt to query ATA functionality via ATA Pass-Through (16) as this page is defined as "must" for SATL devices, hence indicating that the device is at least likely to support Pass-Through (16). This eliminates errors produced by CTL when ATA Pass-Through (16) fails. Switch ATA probe daerror call to SF_NO_PRINT to avoid errors printing out for devices which return invalid errors. Output details about supported and choosen delete method when verbose booted. Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 14:37:23 +00:00
static void daprobedone(struct cam_periph *periph, union ccb *ccb);
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
static periph_ctor_t daregister;
static periph_dtor_t dacleanup;
static periph_start_t dastart;
Fix a problem with the way we handled device invalidation when attaching to a device failed. In theory, the same steps that happen when we get an AC_LOST_DEVICE async notification should have been taken when a driver fails to attach. In practice, that wasn't the case. This only affected the da, cd and ch drivers, but the fix affects all peripheral drivers. There were several possible problems: - In the da driver, we didn't remove the peripheral's softc from the da driver's linked list of softcs. Once the peripheral and softc got removed, we'd get a kernel panic the next time the timeout routine called dasendorderedtag(). - In the da, cd and possibly ch drivers, we didn't remove the peripheral's devstat structure from the devstat queue. Once the peripheral and softc were removed, this could cause a panic if anyone tried to access device statistics. (one component of the linked list wouldn't exist anymore) - In the cd driver, we didn't take the peripheral off the changer run queue if it was scheduled to run. In practice, it's highly unlikely, and maybe impossible that the peripheral would have been on the changer run queue at that stage of the probe process. The fix is: - Add a new peripheral callback function (the "oninvalidate" function) that is called the first time cam_periph_invalidate() is called for a peripheral. - Create new foooninvalidate() routines for each peripheral driver. This routine is always called at splsoftcam(), and contains all the stuff that used to be in the AC_LOST_DEVICE case of the async callback handler. - Move the devstat cleanup call to the destructor/cleanup routines, since some of the drivers do I/O in their close routines. - Make sure that when we're flushing the buffer queue, we traverse it at splbio(). - Add a check for the invalid flag in the pt driver's open routine. Reviewed by: gibbs
1998-10-22 22:16:56 +00:00
static periph_oninv_t daoninvalidate;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
static void dadone(struct cam_periph *periph,
union ccb *done_ccb);
static int daerror(union ccb *ccb, u_int32_t cam_flags,
u_int32_t sense_flags);
static void daprevent(struct cam_periph *periph, int action);
Refactor da(4) to remove one of two code paths used to query capacity data. Reviewed by: ken, mav (earlier version) Sponsored by: The FreeBSD Foundation
2012-04-12 12:58:14 +00:00
static void dareprobe(struct cam_periph *periph);
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
static void dasetgeom(struct cam_periph *periph, uint32_t block_len,
Add CAM infrastructure to allow reporting when a drive's long read capacity data changes. cam_ccb.h: Add a new advanced information type, CDAI_TYPE_RCAPLONG, for long read capacity data. cam_xpt_internal.h: Add a read capacity data pointer and length to struct cam_ed. cam_xpt.c: Free the read capacity buffer when a device goes away. While we're here, make sure we don't leak memory for other malloced fields in struct cam_ed. scsi_all.c: Update the scsi_read_capacity_16() to take a uint8_t * and a length instead of just a pointer to the parameter data structure. This will hopefully make this function somewhat immune to future changes in the parameter data. scsi_all.h: Add some extra bit definitions to struct scsi_read_capacity_data_long, and bump up the structure size to the full size specified by SBC-3. Change the prototype for scsi_read_capacity_16(). scsi_da.c: Register changes in read capacity data with the transport layer. This allows the transport layer to send out an async notification to interested parties. Update the dasetgeom() API. Use scsi_extract_sense_len() instead of scsi_extract_sense(). scsi_xpt.c: Add support for the new CDAI_TYPE_RCAPLONG advanced information type. Make sure we set the physpath pointer to NULL after freeing it. This allows blindly freeing it in the struct cam_ed destructor. sys/param.h: Bump __FreeBSD_version from 1000005 to 1000006 to make it easier for third party drivers to determine that the read capacity data async notification is available. camcontrol.c, mptutil/mpt_cam.c: Update these for the new scsi_read_capacity_16() argument structure. Sponsored by: Spectra Logic
2012-01-26 18:09:28 +00:00
uint64_t maxsector,
struct scsi_read_capacity_data_long *rcaplong,
size_t rcap_size);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
static timeout_t dasendorderedtag;
Implement a new generic mechanism for attaching handler functions to events, in order to pave the way for removing a number of the ad-hoc implementations currently in use. Retire the at_shutdown family of functions and replace them with new event handler lists. Rework kern_shutdown.c to take greater advantage of the use of event handlers. Reviewed by: green
1999-08-21 06:24:40 +00:00
static void dashutdown(void *arg, int howto);
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
static timeout_t damediapoll;
#ifndef DA_DEFAULT_POLL_PERIOD
#define DA_DEFAULT_POLL_PERIOD 3
#endif
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
#ifndef DA_DEFAULT_TIMEOUT
#define DA_DEFAULT_TIMEOUT 60 /* Timeout in seconds */
#endif
Add SYSCTL ints for default normal I/O timeout && retry counts. This is useful if you want to dynamically move into a Fibre Channel or Multi-initiator environment that happens to be particularly noisy and ugly that requires a lot of retries (with shorter I/O timeouts) for commands destried by LIPs or Bus Resets. Reviewed by: deafening silence on audit && scsi on the retry counts MFC after: 2 weeks
2001-07-09 19:18:00 +00:00
#ifndef DA_DEFAULT_RETRY
#define DA_DEFAULT_RETRY 4
#endif
Add a tunable that allows one to turn off the automatic sending of the ORDERED tag. This recoups significant performance gains for many arrays. The default is still to send out the ORDERED tag periodically. Reviewed by: scsi (justin+timeout)
2006-11-02 21:12:37 +00:00
#ifndef DA_DEFAULT_SEND_ORDERED
#define DA_DEFAULT_SEND_ORDERED 1
#endif
Adds the ability to enable / disable sorting of BIO requests queued within CAM. This can significantly improve performance particularly for SSDs which don't suffer from seek latencies. The sysctl / tunable kern.cam.sort_io_queues provides the systems default setting where:- 0 = queued BIOs are NOT sorted 1 = queued BIOs are sorted (default) Each device gets its own sysctl kern.cam.<type>.<id>.sort_io_queue Valid values are:- -1 = use system default (default) 0 = queued BIOs are NOT sorted 1 = queued BIOs are sorted Note: Additional patch will look to add automatic use of none sorted queues for none rotating media e.g. SSD's Reviewed by: scottl Approved by: pjd (mentor) MFC after: 2 weeks
2013-03-29 22:58:15 +00:00
#define DA_SIO (softc->sort_io_queue >= 0 ? \
softc->sort_io_queue : cam_sort_io_queues)
Add a tunable that allows one to turn off the automatic sending of the ORDERED tag. This recoups significant performance gains for many arrays. The default is still to send out the ORDERED tag periodically. Reviewed by: scsi (justin+timeout)
2006-11-02 21:12:37 +00:00
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
static int da_poll_period = DA_DEFAULT_POLL_PERIOD;
Add SYSCTL ints for default normal I/O timeout && retry counts. This is useful if you want to dynamically move into a Fibre Channel or Multi-initiator environment that happens to be particularly noisy and ugly that requires a lot of retries (with shorter I/O timeouts) for commands destried by LIPs or Bus Resets. Reviewed by: deafening silence on audit && scsi on the retry counts MFC after: 2 weeks
2001-07-09 19:18:00 +00:00
static int da_retry_count = DA_DEFAULT_RETRY;
static int da_default_timeout = DA_DEFAULT_TIMEOUT;
Add a tunable that allows one to turn off the automatic sending of the ORDERED tag. This recoups significant performance gains for many arrays. The default is still to send out the ORDERED tag periodically. Reviewed by: scsi (justin+timeout)
2006-11-02 21:12:37 +00:00
static int da_send_ordered = DA_DEFAULT_SEND_ORDERED;
Add SYSCTL ints for default normal I/O timeout && retry counts. This is useful if you want to dynamically move into a Fibre Channel or Multi-initiator environment that happens to be particularly noisy and ugly that requires a lot of retries (with shorter I/O timeouts) for commands destried by LIPs or Bus Resets. Reviewed by: deafening silence on audit && scsi on the retry counts MFC after: 2 weeks
2001-07-09 19:18:00 +00:00
Mark all SYSCTL_NODEs static that have no corresponding SYSCTL_DECLs. The SYSCTL_NODE macro defines a list that stores all child-elements of that node. If there's no SYSCTL_DECL macro anywhere else, there's no reason why it shouldn't be static.
2011-11-07 15:43:11 +00:00
static SYSCTL_NODE(_kern_cam, OID_AUTO, da, CTLFLAG_RD, 0,
Add SYSCTL ints for default normal I/O timeout && retry counts. This is useful if you want to dynamically move into a Fibre Channel or Multi-initiator environment that happens to be particularly noisy and ugly that requires a lot of retries (with shorter I/O timeouts) for commands destried by LIPs or Bus Resets. Reviewed by: deafening silence on audit && scsi on the retry counts MFC after: 2 weeks
2001-07-09 19:18:00 +00:00
"CAM Direct Access Disk driver");
Pull in r267961 and r267973 again. Fix for issues reported will follow.
2014-06-28 03:56:17 +00:00
SYSCTL_INT(_kern_cam_da, OID_AUTO, poll_period, CTLFLAG_RWTUN,
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
&da_poll_period, 0, "Media polling period in seconds");
Pull in r267961 and r267973 again. Fix for issues reported will follow.
2014-06-28 03:56:17 +00:00
SYSCTL_INT(_kern_cam_da, OID_AUTO, retry_count, CTLFLAG_RWTUN,
Add SYSCTL ints for default normal I/O timeout && retry counts. This is useful if you want to dynamically move into a Fibre Channel or Multi-initiator environment that happens to be particularly noisy and ugly that requires a lot of retries (with shorter I/O timeouts) for commands destried by LIPs or Bus Resets. Reviewed by: deafening silence on audit && scsi on the retry counts MFC after: 2 weeks
2001-07-09 19:18:00 +00:00
&da_retry_count, 0, "Normal I/O retry count");
Pull in r267961 and r267973 again. Fix for issues reported will follow.
2014-06-28 03:56:17 +00:00
SYSCTL_INT(_kern_cam_da, OID_AUTO, default_timeout, CTLFLAG_RWTUN,
Add SYSCTL ints for default normal I/O timeout && retry counts. This is useful if you want to dynamically move into a Fibre Channel or Multi-initiator environment that happens to be particularly noisy and ugly that requires a lot of retries (with shorter I/O timeouts) for commands destried by LIPs or Bus Resets. Reviewed by: deafening silence on audit && scsi on the retry counts MFC after: 2 weeks
2001-07-09 19:18:00 +00:00
&da_default_timeout, 0, "Normal I/O timeout (in seconds)");
Pull in r267961 and r267973 again. Fix for issues reported will follow.
2014-06-28 03:56:17 +00:00
SYSCTL_INT(_kern_cam_da, OID_AUTO, send_ordered, CTLFLAG_RWTUN,
Add a tunable that allows one to turn off the automatic sending of the ORDERED tag. This recoups significant performance gains for many arrays. The default is still to send out the ORDERED tag periodically. Reviewed by: scsi (justin+timeout)
2006-11-02 21:12:37 +00:00
&da_send_ordered, 0, "Send Ordered Tags");
Add SYSCTL ints for default normal I/O timeout && retry counts. This is useful if you want to dynamically move into a Fibre Channel or Multi-initiator environment that happens to be particularly noisy and ugly that requires a lot of retries (with shorter I/O timeouts) for commands destried by LIPs or Bus Resets. Reviewed by: deafening silence on audit && scsi on the retry counts MFC after: 2 weeks
2001-07-09 19:18:00 +00:00
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
/*
* DA_ORDEREDTAG_INTERVAL determines how often, relative
* to the default timeout, we check to see whether an ordered
* tagged transaction is appropriate to prevent simple tag
* starvation. Since we'd like to ensure that there is at least
* 1/2 of the timeout length left for a starved transaction to
* complete after we've sent an ordered tag, we must poll at least
* four times in every timeout period. This takes care of the worst
* case where a starved transaction starts during an interval that
* meets the requirement "don't send an ordered tag" test so it takes
* us two intervals to determine that a tag must be sent.
*/
#ifndef DA_ORDEREDTAG_INTERVAL
#define DA_ORDEREDTAG_INTERVAL 4
#endif
static struct periph_driver dadriver =
{
dainit, "da",
TAILQ_HEAD_INITIALIZER(dadriver.units), /* generation */ 0
};
Change the peripheral driver list from a linker set to module driven driver registration. This should allow things like da, sa, cd etc to be in seperate KLD's to the cam core and make them preloadable.
2001-02-07 07:05:59 +00:00
PERIPHDRIVER_DECLARE(da, dadriver);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
Mark MALLOC_DEFINEs static that have no corresponding MALLOC_DECLAREs. This means that their use is restricted to a single C file.
2011-11-07 06:44:47 +00:00
static MALLOC_DEFINE(M_SCSIDA, "scsi_da", "scsi_da buffers");
Eliminate the use of M_TEMP.
2007-05-14 21:48:53 +00:00
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
static int
NO_GEOM cleanup: Move to "struct disk *" centric api.
2003-02-25 22:06:21 +00:00
daopen(struct disk *dp)
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
{
struct cam_periph *periph;
struct da_softc *softc;
int error;
NO_GEOM cleanup: Move to "struct disk *" centric api.
2003-02-25 22:06:21 +00:00
periph = (struct cam_periph *)dp->d_drv1;
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
if (cam_periph_acquire(periph) != CAM_REQ_CMP) {
Fix a race condition in CAM peripheral free handling, locking in the CAM XPT bus traversal code, and a number of other periph level issues. cam_periph.h, cam_periph.c: Modify cam_periph_acquire() to test the CAM_PERIPH_INVALID flag prior to allowing a reference count to be gained on a peripheral. Callers of this function will receive CAM_REQ_CMP_ERR status in the situation of attempting to reference an invalidated periph. This guarantees that a peripheral scheduled for a deferred free will not be accessed during its wait for destruction. Panic during attempts to drop a reference count on a peripheral that already has a zero reference count. In cam_periph_list(), use a local sbuf with SBUF_FIXEDLEN set so that mallocs do not occur while the xpt topology lock is held, regardless of the allocation policy of the passed in sbuf. Add a new routine, cam_periph_release_locked_buses(), that can be called when the caller already holds the CAM topology lock. Add some extra debugging for duplicate peripheral allocations in cam_periph_alloc(). Treat CAM_DEV_NOT_THERE much the same as a selection timeout (AC_LOST_DEVICE is emitted), but forgo retries. cam_xpt.c: Revamp the way the EDT traversal code does locking and reference counting. This was broken, since it assumed that the EDT would not change during traversal, but that assumption is no longer valid. So, to prevent devices from going away while we traverse the EDT, make sure we properly lock everything and hold references on devices that we are using. The two peripheral driver traversal routines should be examined. xptpdperiphtraverse() holds the topology lock for the entire time it runs. xptperiphtraverse() is now locked properly, but only holds the topology lock while it is traversing the list, and not while the traversal function is running. The bus locking code in xptbustraverse() should also be revisited at a later time, since it is complex and should probably be simplified. scsi_da.c: Pay attention to the return value from cam_periph_acquire(). Return 0 always from daclose() even if the disk is now gone. Add some rudimentary error injection support. scsi_sg.c: Fix reference counting in the sg(4) driver. The sg driver was calling cam_periph_release() on close, but never called cam_periph_acquire() (which increments the reference count) on open. The periph code correctly complained that the sg(4) driver was trying to decrement the refcount when it was already 0. Sponsored by: Spectra Logic MFC after: 2 weeks
2012-01-12 00:41:48 +00:00
return (ENXIO);
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
}
cam_periph_lock(periph);
if ((error = cam_periph_hold(periph, PRIBIO|PCATCH)) != 0) {
cam_periph_unlock(periph);
cam_periph_release(periph);
return (error);
}
Tune and add some missing CAM_DEBUG() points for better consistency.
2012-06-05 09:45:42 +00:00
CAM_DEBUG(periph->path, CAM_DEBUG_TRACE | CAM_DEBUG_PERIPH,
("daopen\n"));
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
MFprojects/camlock r249542: Remove ADA_FLAG_PACK_INVALID flag. Since ATA disks have no concept of media change it only duplicates CAM_PERIPH_INVALID flag, so we can use last one. Slightly cleanup DA_FLAG_PACK_INVALID use.
2013-04-27 12:46:04 +00:00
softc = (struct da_softc *)periph->softc;
Refactor da(4) to remove one of two code paths used to query capacity data. Reviewed by: ken, mav (earlier version) Sponsored by: The FreeBSD Foundation
2012-04-12 12:58:14 +00:00
dareprobe(periph);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
Refactor da(4) to remove one of two code paths used to query capacity data. Reviewed by: ken, mav (earlier version) Sponsored by: The FreeBSD Foundation
2012-04-12 12:58:14 +00:00
/* Wait for the disk size update. */
MFprojects/camlock r248931: Replace some direct mutex operations with wrappers. MFC after: 2 weeks
2013-04-04 19:07:37 +00:00
error = cam_periph_sleep(periph, &softc->disk->d_mediasize, PRIBIO,
Refactor da(4) to remove one of two code paths used to query capacity data. Reviewed by: ken, mav (earlier version) Sponsored by: The FreeBSD Foundation
2012-04-12 12:58:14 +00:00
"dareprobe", 0);
if (error != 0)
Correct short delete issue in SCSI UNMAP support Correct missing \n's in xpt_print's Correct incorrect count being passed to short delete xpt_print MFC after: 1 week
2014-01-04 17:52:43 +00:00
xpt_print(periph->path, "unable to retrieve capacity data\n");
Some fixes for the CD and DA drivers from bde. (and some of my own as well) Among them: [ cd driver ] 1. Old labeling code was still there. 2. Error handling for dsopen() was broken (no test for the `error' returned by dsopen(); bogus test of an `error' that is known to be 0). 3. cdopen() closed the physical device after certain errors although there may still be open partitions on it. 4. cdclose() closed the physical device although there may still be open partitions on it. 5. Some printf format fixes was incomplete or missing. 6. cdioctl() truncated unit numbers mod 256. 7. cdioctl() was missing locking. [ da driver ] 1. daclose() closed the physical device although there may still be open partitions on it. This was fixed many years ago in sd.c rev.1.57. 2. A minor optimization (the dk_slices != NULL test) in sdopen() became uglier in daopen(). It is not worth doing. da only regressed compared with od and my version of sd, since I never committed the change to sd. daopen() should probably do less if some partition is already open. This is not addressed by the diffs. [ ... ] 5. "opt_hw_wdog.h" was not included, so the HW_WDOG code was unreachable. - Added a getdev CCB call in the cdopen() and daopen() calls so that the vendor name and device name are available for the disklabel. (suggested by bde) - Removed vestigal devfs support in both drivers, since we can't properly work with devfs yet. (ask bde for details on this) - Cleaned up the probe code in both drivers in the failure cases. There were a number of things wrong here. The peripheral driver instances weren't getting properly cleaned up. Sometimes the wrong probe message would get printed out (with the failure message appended), so it wasn't very clear that we failed to attach. SCSI sense information was printed, even when the error in question wasn't a SCSI error. I put similar fixes into the changer driver in revision 1.2 of scsi_ch.c. Reviewed by: gibbs Submitted by: bde (partially)
1998-10-07 02:57:57 +00:00
Tune support for removable media in da driver: - remove DA_FLAG_SAW_MEDIA flag, almost opposite to DA_FLAG_PACK_INVALID, using the last instead. - allow opening device with no media present, reporting zero media size and non-zero sector size, as geom/notes suggests. That allow to read device attributes and potentially do other things, not related to media.
2013-05-03 11:53:06 +00:00
if (periph->flags & CAM_PERIPH_INVALID)
Refactor da(4) to remove one of two code paths used to query capacity data. Reviewed by: ken, mav (earlier version) Sponsored by: The FreeBSD Foundation
2012-04-12 12:58:14 +00:00
error = ENXIO;
Fix a locking mistake in daopen(). If the open fails, which can happen because the media was removed, the periph would get its refcount dropped and ultimately freed before getting unlocked. This created a dangling pointer that was easy to trip over. This fixes a common source of crashes with removaable media, but problems remain and will get tracked down.
2008-08-29 04:39:46 +00:00
Refactor da(4) to remove one of two code paths used to query capacity data. Reviewed by: ken, mav (earlier version) Sponsored by: The FreeBSD Foundation
2012-04-12 12:58:14 +00:00
if (error == 0 && (softc->flags & DA_FLAG_PACK_REMOVABLE) != 0 &&
(softc->quirks & DA_Q_NO_PREVENT) == 0)
daprevent(periph, PR_PREVENT);
Fix a locking mistake in daopen(). If the open fails, which can happen because the media was removed, the periph would get its refcount dropped and ultimately freed before getting unlocked. This created a dangling pointer that was easy to trip over. This fixes a common source of crashes with removaable media, but problems remain and will get tracked down.
2008-08-29 04:39:46 +00:00
Tune support for removable media in da driver: - remove DA_FLAG_SAW_MEDIA flag, almost opposite to DA_FLAG_PACK_INVALID, using the last instead. - allow opening device with no media present, reporting zero media size and non-zero sector size, as geom/notes suggests. That allow to read device attributes and potentially do other things, not related to media.
2013-05-03 11:53:06 +00:00
if (error == 0) {
softc->flags &= ~DA_FLAG_PACK_INVALID;
softc->flags |= DA_FLAG_OPEN;
}
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
cam_periph_unhold(periph);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
cam_periph_unlock(periph);
Fix a locking mistake in daopen(). If the open fails, which can happen because the media was removed, the periph would get its refcount dropped and ultimately freed before getting unlocked. This created a dangling pointer that was easy to trip over. This fixes a common source of crashes with removaable media, but problems remain and will get tracked down.
2008-08-29 04:39:46 +00:00
Tune support for removable media in da driver: - remove DA_FLAG_SAW_MEDIA flag, almost opposite to DA_FLAG_PACK_INVALID, using the last instead. - allow opening device with no media present, reporting zero media size and non-zero sector size, as geom/notes suggests. That allow to read device attributes and potentially do other things, not related to media.
2013-05-03 11:53:06 +00:00
if (error != 0)
Fix a locking mistake in daopen(). If the open fails, which can happen because the media was removed, the periph would get its refcount dropped and ultimately freed before getting unlocked. This created a dangling pointer that was easy to trip over. This fixes a common source of crashes with removaable media, but problems remain and will get tracked down.
2008-08-29 04:39:46 +00:00
cam_periph_release(periph);
Refactor da(4) to remove one of two code paths used to query capacity data. Reviewed by: ken, mav (earlier version) Sponsored by: The FreeBSD Foundation
2012-04-12 12:58:14 +00:00
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
return (error);
}
static int
NO_GEOM cleanup: Move to "struct disk *" centric api.
2003-02-25 22:06:21 +00:00
daclose(struct disk *dp)
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
{
struct cam_periph *periph;
struct da_softc *softc;
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
union ccb *ccb;
Synchronize device cache on close only if there were some write operations. While these operations are not really needed otherwise, at least for SCSI they may cause extra errors if some other initiator holds write exclusive reservation on the LUN (SYNCHRONIZE CACHE handled as "write" operation).
2013-07-27 22:44:55 +00:00
int error;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
NO_GEOM cleanup: Move to "struct disk *" centric api.
2003-02-25 22:06:21 +00:00
periph = (struct cam_periph *)dp->d_drv1;
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
softc = (struct da_softc *)periph->softc;
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
cam_periph_lock(periph);
Tune and add some missing CAM_DEBUG() points for better consistency.
2012-06-05 09:45:42 +00:00
CAM_DEBUG(periph->path, CAM_DEBUG_TRACE | CAM_DEBUG_PERIPH,
("daclose\n"));
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
if (cam_periph_hold(periph, PRIBIO) == 0) {
/* Flush disk cache. */
if ((softc->flags & DA_FLAG_DIRTY) != 0 &&
(softc->quirks & DA_Q_NO_SYNC_CACHE) == 0 &&
(softc->flags & DA_FLAG_PACK_INVALID) == 0) {
ccb = cam_periph_getccb(periph, CAM_PRIORITY_NORMAL);
scsi_synchronize_cache(&ccb->csio, /*retries*/1,
/*cbfcnp*/dadone, MSG_SIMPLE_Q_TAG,
/*begin_lba*/0, /*lb_count*/0, SSD_FULL_SIZE,
5 * 60 * 1000);
error = cam_periph_runccb(ccb, daerror, /*cam_flags*/0,
/*sense_flags*/SF_RETRY_UA | SF_QUIET_IR,
softc->disk->d_devstat);
if (error == 0)
softc->flags &= ~DA_FLAG_DIRTY;
xpt_release_ccb(ccb);
}
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
/* Allow medium removal. */
if ((softc->flags & DA_FLAG_PACK_REMOVABLE) != 0 &&
(softc->quirks & DA_Q_NO_PREVENT) == 0)
daprevent(periph, PR_ALLOW);
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
cam_periph_unhold(periph);
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
}
Don't leave the device queue in a frozen state if the Synchronize Cache command on close fails.
1998-09-19 04:59:35 +00:00
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
/*
* If we've got removeable media, mark the blocksize as
* unavailable, since it could change when new media is
* inserted.
*/
if ((softc->flags & DA_FLAG_PACK_REMOVABLE) != 0)
Change the disk(9) API in order to make device removal more robust. Previously the "struct disk" were owned by the device driver and this gave us problems when the device disappared and the users of that device were not immediately disappearing. Now the struct disk is allocate with a new call, disk_alloc() and owned by geom_disk and just abandonned by the device driver when disk_create() is called. Unfortunately, this results in a ton of "s/\./->/" changes to device drivers. Since I'm doing the sweep anyway, a couple of other API improvements have been carried out at the same time: The Giant awareness flag has been flipped from DISKFLAG_NOGIANT to DISKFLAG_NEEDSGIANT A version number have been added to disk_create() so that we can detect, report and ignore binary drivers with old ABI in the future. Manual page update to follow shortly.
2004-02-18 21:36:53 +00:00
softc->disk->d_devstat->flags |= DEVSTAT_BS_UNAVAILABLE;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
softc->flags &= ~DA_FLAG_OPEN;
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
while (softc->refcount != 0)
cam_periph_sleep(periph, &softc->refcount, PRIBIO, "daclose", 1);
Revert r186186 for now; it breaks stuff. Approved by: rwatson (mentor)
2008-12-17 10:49:03 +00:00
cam_periph_unlock(periph);
Periph driver fixes, second try. Reviewed by: scottl Approved by: rwatson (mentor) Sponsored by: FreeBSD Foundation
2008-12-19 14:31:40 +00:00
cam_periph_release(periph);
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
return (0);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
static void
daschedule(struct cam_periph *periph)
{
struct da_softc *softc = (struct da_softc *)periph->softc;
Do not schedule periph for payload/TUR requests if reprobe is in progress to avoid sending extra READ CAPACITY requests by dastart(). Schedule periph again on reprobe completion, or otherwise it may stuck indefinitely long. This should fix USB explore thread hanging on device unplug, waiting for periph destruction. Reported by: hselasky
2013-01-11 16:10:11 +00:00
if (softc->state != DA_STATE_NORMAL)
return;
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
/* Check if we have more work to do. */
if (bioq_first(&softc->bio_queue) ||
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
(!softc->delete_running && bioq_first(&softc->delete_queue)) ||
softc->tur) {
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
xpt_schedule(periph, CAM_PRIORITY_NORMAL);
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
}
}
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
/*
* Actually translate the requested transfer into one the physical driver
* can understand. The transfer is described by a buf and will include
* only one physical transfer.
*/
static void
Complete the bio/buf divorce for all code below devfs::strategy Exceptions: Vinum untouched. This means that it cannot be compiled. Greg Lehey is on the case. CCD not converted yet, casts to struct buf (still safe) atapi-cd casts to struct buf to examine B_PHYS
2000-04-15 05:54:02 +00:00
dastrategy(struct bio *bp)
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
{
struct cam_periph *periph;
struct da_softc *softc;
NO_GEOM cleanup: Move to "struct disk *" centric api.
2003-02-25 22:06:21 +00:00
periph = (struct cam_periph *)bp->bio_disk->d_drv1;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
softc = (struct da_softc *)periph->softc;
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
cam_periph_lock(periph);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
/*
* If the device has been made invalid, error out
*/
if ((softc->flags & DA_FLAG_PACK_INVALID)) {
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
cam_periph_unlock(periph);
Exploit recent improvements in the disk minilayer to simplify error handling a bit. Dogmatic lingupurists can celebrate that a number of gotos got removed. Reviewed by: mjacob, ken
2001-05-08 08:30:48 +00:00
biofinish(bp, NULL, ENXIO);
return;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
Tune and add some missing CAM_DEBUG() points for better consistency.
2012-06-05 09:45:42 +00:00
CAM_DEBUG(periph->path, CAM_DEBUG_TRACE, ("dastrategy(%p)\n", bp));
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
/*
* Place it in the queue of disk activities for this disk
Work around the other uses of M_WAITOK.
2007-06-16 18:20:29 +00:00
*/
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
if (bp->bio_cmd == BIO_DELETE) {
The code that combines adjacent ranges for BIO_DELETEs to optimize trims to the device assumes the list is sorted. Don't apply the optimization of not sorting the queue when we have SSDs to the delete_queue, since it causes more discard traffic to the drive. While one could argue that the higher levels should coalesce the trims, that's not done today, so some optimization at this level is needed. CR: https://phabric.freebsd.org/D142
2014-06-05 17:13:42 +00:00
bioq_disksort(&softc->delete_queue, bp);
Adds the ability to enable / disable sorting of BIO requests queued within CAM. This can significantly improve performance particularly for SSDs which don't suffer from seek latencies. The sysctl / tunable kern.cam.sort_io_queues provides the systems default setting where:- 0 = queued BIOs are NOT sorted 1 = queued BIOs are sorted (default) Each device gets its own sysctl kern.cam.<type>.<id>.sort_io_queue Valid values are:- -1 = use system default (default) 0 = queued BIOs are NOT sorted 1 = queued BIOs are sorted Note: Additional patch will look to add automatic use of none sorted queues for none rotating media e.g. SSD's Reviewed by: scottl Approved by: pjd (mentor) MFC after: 2 weeks
2013-03-29 22:58:15 +00:00
} else if (DA_SIO) {
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
bioq_disksort(&softc->bio_queue, bp);
Adds the ability to enable / disable sorting of BIO requests queued within CAM. This can significantly improve performance particularly for SSDs which don't suffer from seek latencies. The sysctl / tunable kern.cam.sort_io_queues provides the systems default setting where:- 0 = queued BIOs are NOT sorted 1 = queued BIOs are sorted (default) Each device gets its own sysctl kern.cam.<type>.<id>.sort_io_queue Valid values are:- -1 = use system default (default) 0 = queued BIOs are NOT sorted 1 = queued BIOs are sorted Note: Additional patch will look to add automatic use of none sorted queues for none rotating media e.g. SSD's Reviewed by: scottl Approved by: pjd (mentor) MFC after: 2 weeks
2013-03-29 22:58:15 +00:00
} else {
bioq_insert_tail(&softc->bio_queue, bp);
}
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
/*
* Schedule ourselves for performing the work.
*/
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
daschedule(periph);
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
cam_periph_unlock(periph);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
return;
}
static int
NO_GEOM cleanup: Retire the "d_dump_t" and use the "dumper_t" type instead. Dumper_t takes a void * as first arg which is more general than the dev_t taken by d_dump_t. (Remember: we could have net-dumpers if somebody wrote us one!) Define the convention for GEOM controlled disk devices to be that the first argument to the dumper function is the struct disk pointer. Change device drivers accordingly.
2003-02-21 19:00:48 +00:00
dadump(void *arg, void *virtual, vm_offset_t physical, off_t offset, size_t length)
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
{
Convert dadump to use reasonable data types so that some casting is unecessary.
1998-12-11 03:54:43 +00:00
struct cam_periph *periph;
struct da_softc *softc;
Introduce the disk mini-layer and devstat_end_transaction_buf() in cam/scsi. Somewhat reviewed by: ken
1999-10-01 09:34:10 +00:00
u_int secsize;
Convert dadump to use reasonable data types so that some casting is unecessary.
1998-12-11 03:54:43 +00:00
struct ccb_scsiio csio;
NO_GEOM cleanup: Retire the "d_dump_t" and use the "dumper_t" type instead. Dumper_t takes a void * as first arg which is more general than the dev_t taken by d_dump_t. (Remember: we could have net-dumpers if somebody wrote us one!) Define the convention for GEOM controlled disk devices to be that the first argument to the dumper function is the struct disk pointer. Change device drivers accordingly.
2003-02-21 19:00:48 +00:00
struct disk *dp;
One more major cam_periph_error() rewrite to improve error handling and reporting. It includes: - removing of error messages controlled by bootverbose, replacing them with more universal and informative debugging on CAM_DEBUG_INFO level, that is now built into the kernel by default; - more close following to the arguments submitted by caller, such as SF_PRINT_ALWAYS, SF_QUIET_IR and SF_NO_PRINT; consumer knows better which errors are usual/expected at this point and which are really informative; - adding two new flags SF_NO_RECOVERY and SF_NO_RETRY to allow caller specify how much assistance it needs at this point; previously consumers controlled that by not calling cam_periph_error() at all, but that made behavior inconsistent and debugging complicated; - tuning debug messages and taken actions order to make debugging output more readable and cause-effect relationships visible; - making camperiphdone() (common device recovery completion handler) to also use cam_periph_error() in most cases, instead of own dumb code; - removing manual sense fetching code from cam_periph_error(); I was told by number of people that it is SIM obligation to fetch sense data, so this code is useless and only significantly complicates recovery logic; - making ada, da and pass driver to use cam_periph_error() with new limited recovery options to handle error recovery and debugging in common way; as one of results, CAM_REQUEUE_REQ and other retrying statuses are now working fine with pass driver, that caused many problems before. - reverting r186891 by raj@ to avoid burning few seconds in tight DELAY() loops on device probe, while device simply loads media; I think that problem may already be fixed in other way, and even if it is not, solution must be different. Sponsored by: iXsystems, Inc. MFC after: 2 weeks
2012-06-09 13:07:44 +00:00
int error = 0;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
NO_GEOM cleanup: Retire the "d_dump_t" and use the "dumper_t" type instead. Dumper_t takes a void * as first arg which is more general than the dev_t taken by d_dump_t. (Remember: we could have net-dumpers if somebody wrote us one!) Define the convention for GEOM controlled disk devices to be that the first argument to the dumper function is the struct disk pointer. Change device drivers accordingly.
2003-02-21 19:00:48 +00:00
dp = arg;
NO_GEOM cleanup: Move to "struct disk *" centric api.
2003-02-25 22:06:21 +00:00
periph = dp->d_drv1;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
softc = (struct da_softc *)periph->softc;
Missed locking the dump and shutdown entry points in the scsi_da driver.
2007-04-18 05:14:16 +00:00
cam_periph_lock(periph);
DA (scsi) and AD (ata) diskdrivers: Make the dump routine do just writing of data.
2002-03-31 22:28:03 +00:00
secsize = softc->params.secsize;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
Missed locking the dump and shutdown entry points in the scsi_da driver.
2007-04-18 05:14:16 +00:00
if ((softc->flags & DA_FLAG_PACK_INVALID) != 0) {
cam_periph_unlock(periph);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
return (ENXIO);
Missed locking the dump and shutdown entry points in the scsi_da driver.
2007-04-18 05:14:16 +00:00
}
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
DA (scsi) and AD (ata) diskdrivers: Make the dump routine do just writing of data.
2002-03-31 22:28:03 +00:00
if (length > 0) {
Replace most of priority numbers with defines. No logical changes.
2009-10-23 08:27:55 +00:00
xpt_setup_ccb(&csio.ccb_h, periph->path, CAM_PRIORITY_NORMAL);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
csio.ccb_h.ccb_state = DA_CCB_DUMP;
scsi_read_write(&csio,
One more major cam_periph_error() rewrite to improve error handling and reporting. It includes: - removing of error messages controlled by bootverbose, replacing them with more universal and informative debugging on CAM_DEBUG_INFO level, that is now built into the kernel by default; - more close following to the arguments submitted by caller, such as SF_PRINT_ALWAYS, SF_QUIET_IR and SF_NO_PRINT; consumer knows better which errors are usual/expected at this point and which are really informative; - adding two new flags SF_NO_RECOVERY and SF_NO_RETRY to allow caller specify how much assistance it needs at this point; previously consumers controlled that by not calling cam_periph_error() at all, but that made behavior inconsistent and debugging complicated; - tuning debug messages and taken actions order to make debugging output more readable and cause-effect relationships visible; - making camperiphdone() (common device recovery completion handler) to also use cam_periph_error() in most cases, instead of own dumb code; - removing manual sense fetching code from cam_periph_error(); I was told by number of people that it is SIM obligation to fetch sense data, so this code is useless and only significantly complicates recovery logic; - making ada, da and pass driver to use cam_periph_error() with new limited recovery options to handle error recovery and debugging in common way; as one of results, CAM_REQUEUE_REQ and other retrying statuses are now working fine with pass driver, that caused many problems before. - reverting r186891 by raj@ to avoid burning few seconds in tight DELAY() loops on device probe, while device simply loads media; I think that problem may already be fixed in other way, and even if it is not, solution must be different. Sponsored by: iXsystems, Inc. MFC after: 2 weeks
2012-06-09 13:07:44 +00:00
/*retries*/0,
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
dadone,
MSG_ORDERED_Q_TAG,
Support unmapped i/o for the md(4). The vnode-backed md(4) has to map the unmapped bio because VOP_READ() and VOP_WRITE() interfaces do not allow to pass unmapped requests to the filesystem. Vnode-backed md(4) uses pbufs instead of relying on the bio_transient_map, to avoid usual md deadlock. Sponsored by: The FreeBSD Foundation Tested by: pho, scottl
2013-03-19 15:01:50 +00:00
/*read*/SCSI_RW_WRITE,
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
/*byte2*/0,
"Fix" a problem with the Quantum Viking. It appears that this drive does not like the 6-byte read and write commands! It returns illegal request, with the field pointer pointing to byte 9 of a 6 byte CDB. In any case, the work around is to put in a quirk mechanism that makes sure that we don't send 6-byte reads or writes to this device. It's rather sad that this is necessary. You'd think that they would be able to get something that basic to work right in their firmware... Reviewed by: gibbs Reported by: Adam McDougall <bsdx@spawnet.com>
1998-12-02 17:35:28 +00:00
/*minimum_cmd_size*/ softc->minimum_cmd_size,
DA (scsi) and AD (ata) diskdrivers: Make the dump routine do just writing of data.
2002-03-31 22:28:03 +00:00
offset / secsize,
length / secsize,
/*data_ptr*/(u_int8_t *) virtual,
/*dxfer_len*/length,
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
/*sense_len*/SSD_FULL_SIZE,
Insert ordered command every 1/4 of the current command timeout, not 1/4 of the default one. Without this change setting kern.cam.ada.default_timeout to 1 instead of 30 allowed me to trigger several false positive command timeouts under heavy ZFS load on a SiI3132 siis(4) controller with 5 HDDs on a port multiplier. MFC after: 1 week
2012-02-02 19:02:15 +00:00
da_default_timeout * 1000);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
xpt_polled_action((union ccb *)&csio);
One more major cam_periph_error() rewrite to improve error handling and reporting. It includes: - removing of error messages controlled by bootverbose, replacing them with more universal and informative debugging on CAM_DEBUG_INFO level, that is now built into the kernel by default; - more close following to the arguments submitted by caller, such as SF_PRINT_ALWAYS, SF_QUIET_IR and SF_NO_PRINT; consumer knows better which errors are usual/expected at this point and which are really informative; - adding two new flags SF_NO_RECOVERY and SF_NO_RETRY to allow caller specify how much assistance it needs at this point; previously consumers controlled that by not calling cam_periph_error() at all, but that made behavior inconsistent and debugging complicated; - tuning debug messages and taken actions order to make debugging output more readable and cause-effect relationships visible; - making camperiphdone() (common device recovery completion handler) to also use cam_periph_error() in most cases, instead of own dumb code; - removing manual sense fetching code from cam_periph_error(); I was told by number of people that it is SIM obligation to fetch sense data, so this code is useless and only significantly complicates recovery logic; - making ada, da and pass driver to use cam_periph_error() with new limited recovery options to handle error recovery and debugging in common way; as one of results, CAM_REQUEUE_REQ and other retrying statuses are now working fine with pass driver, that caused many problems before. - reverting r186891 by raj@ to avoid burning few seconds in tight DELAY() loops on device probe, while device simply loads media; I think that problem may already be fixed in other way, and even if it is not, solution must be different. Sponsored by: iXsystems, Inc. MFC after: 2 weeks
2012-06-09 13:07:44 +00:00
error = cam_periph_error((union ccb *)&csio,
0, SF_NO_RECOVERY | SF_NO_RETRY, NULL);
if ((csio.ccb_h.status & CAM_DEV_QFRZN) != 0)
cam_release_devq(csio.ccb_h.path, /*relsim_flags*/0,
/*reduction*/0, /*timeout*/0, /*getcount_only*/0);
if (error != 0)
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
printf("Aborting dump due to I/O error.\n");
One more major cam_periph_error() rewrite to improve error handling and reporting. It includes: - removing of error messages controlled by bootverbose, replacing them with more universal and informative debugging on CAM_DEBUG_INFO level, that is now built into the kernel by default; - more close following to the arguments submitted by caller, such as SF_PRINT_ALWAYS, SF_QUIET_IR and SF_NO_PRINT; consumer knows better which errors are usual/expected at this point and which are really informative; - adding two new flags SF_NO_RECOVERY and SF_NO_RETRY to allow caller specify how much assistance it needs at this point; previously consumers controlled that by not calling cam_periph_error() at all, but that made behavior inconsistent and debugging complicated; - tuning debug messages and taken actions order to make debugging output more readable and cause-effect relationships visible; - making camperiphdone() (common device recovery completion handler) to also use cam_periph_error() in most cases, instead of own dumb code; - removing manual sense fetching code from cam_periph_error(); I was told by number of people that it is SIM obligation to fetch sense data, so this code is useless and only significantly complicates recovery logic; - making ada, da and pass driver to use cam_periph_error() with new limited recovery options to handle error recovery and debugging in common way; as one of results, CAM_REQUEUE_REQ and other retrying statuses are now working fine with pass driver, that caused many problems before. - reverting r186891 by raj@ to avoid burning few seconds in tight DELAY() loops on device probe, while device simply loads media; I think that problem may already be fixed in other way, and even if it is not, solution must be different. Sponsored by: iXsystems, Inc. MFC after: 2 weeks
2012-06-09 13:07:44 +00:00
cam_periph_unlock(periph);
return (error);
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
}
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
/*
* Sync the disk cache contents to the physical media.
*/
if ((softc->quirks & DA_Q_NO_SYNC_CACHE) == 0) {
Replace most of priority numbers with defines. No logical changes.
2009-10-23 08:27:55 +00:00
xpt_setup_ccb(&csio.ccb_h, periph->path, CAM_PRIORITY_NORMAL);
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
csio.ccb_h.ccb_state = DA_CCB_DUMP;
scsi_synchronize_cache(&csio,
One more major cam_periph_error() rewrite to improve error handling and reporting. It includes: - removing of error messages controlled by bootverbose, replacing them with more universal and informative debugging on CAM_DEBUG_INFO level, that is now built into the kernel by default; - more close following to the arguments submitted by caller, such as SF_PRINT_ALWAYS, SF_QUIET_IR and SF_NO_PRINT; consumer knows better which errors are usual/expected at this point and which are really informative; - adding two new flags SF_NO_RECOVERY and SF_NO_RETRY to allow caller specify how much assistance it needs at this point; previously consumers controlled that by not calling cam_periph_error() at all, but that made behavior inconsistent and debugging complicated; - tuning debug messages and taken actions order to make debugging output more readable and cause-effect relationships visible; - making camperiphdone() (common device recovery completion handler) to also use cam_periph_error() in most cases, instead of own dumb code; - removing manual sense fetching code from cam_periph_error(); I was told by number of people that it is SIM obligation to fetch sense data, so this code is useless and only significantly complicates recovery logic; - making ada, da and pass driver to use cam_periph_error() with new limited recovery options to handle error recovery and debugging in common way; as one of results, CAM_REQUEUE_REQ and other retrying statuses are now working fine with pass driver, that caused many problems before. - reverting r186891 by raj@ to avoid burning few seconds in tight DELAY() loops on device probe, while device simply loads media; I think that problem may already be fixed in other way, and even if it is not, solution must be different. Sponsored by: iXsystems, Inc. MFC after: 2 weeks
2012-06-09 13:07:44 +00:00
/*retries*/0,
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
/*cbfcnp*/dadone,
MSG_SIMPLE_Q_TAG,
/*begin_lba*/0,/* Cover the whole disk */
/*lb_count*/0,
SSD_FULL_SIZE,
5 * 60 * 1000);
xpt_polled_action((union ccb *)&csio);
One more major cam_periph_error() rewrite to improve error handling and reporting. It includes: - removing of error messages controlled by bootverbose, replacing them with more universal and informative debugging on CAM_DEBUG_INFO level, that is now built into the kernel by default; - more close following to the arguments submitted by caller, such as SF_PRINT_ALWAYS, SF_QUIET_IR and SF_NO_PRINT; consumer knows better which errors are usual/expected at this point and which are really informative; - adding two new flags SF_NO_RECOVERY and SF_NO_RETRY to allow caller specify how much assistance it needs at this point; previously consumers controlled that by not calling cam_periph_error() at all, but that made behavior inconsistent and debugging complicated; - tuning debug messages and taken actions order to make debugging output more readable and cause-effect relationships visible; - making camperiphdone() (common device recovery completion handler) to also use cam_periph_error() in most cases, instead of own dumb code; - removing manual sense fetching code from cam_periph_error(); I was told by number of people that it is SIM obligation to fetch sense data, so this code is useless and only significantly complicates recovery logic; - making ada, da and pass driver to use cam_periph_error() with new limited recovery options to handle error recovery and debugging in common way; as one of results, CAM_REQUEUE_REQ and other retrying statuses are now working fine with pass driver, that caused many problems before. - reverting r186891 by raj@ to avoid burning few seconds in tight DELAY() loops on device probe, while device simply loads media; I think that problem may already be fixed in other way, and even if it is not, solution must be different. Sponsored by: iXsystems, Inc. MFC after: 2 weeks
2012-06-09 13:07:44 +00:00
error = cam_periph_error((union ccb *)&csio,
0, SF_NO_RECOVERY | SF_NO_RETRY | SF_QUIET_IR, NULL);
if ((csio.ccb_h.status & CAM_DEV_QFRZN) != 0)
cam_release_devq(csio.ccb_h.path, /*relsim_flags*/0,
/*reduction*/0, /*timeout*/0, /*getcount_only*/0);
if (error != 0)
xpt_print(periph->path, "Synchronize cache failed\n");
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
}
Missed locking the dump and shutdown entry points in the scsi_da driver.
2007-04-18 05:14:16 +00:00
cam_periph_unlock(periph);
One more major cam_periph_error() rewrite to improve error handling and reporting. It includes: - removing of error messages controlled by bootverbose, replacing them with more universal and informative debugging on CAM_DEBUG_INFO level, that is now built into the kernel by default; - more close following to the arguments submitted by caller, such as SF_PRINT_ALWAYS, SF_QUIET_IR and SF_NO_PRINT; consumer knows better which errors are usual/expected at this point and which are really informative; - adding two new flags SF_NO_RECOVERY and SF_NO_RETRY to allow caller specify how much assistance it needs at this point; previously consumers controlled that by not calling cam_periph_error() at all, but that made behavior inconsistent and debugging complicated; - tuning debug messages and taken actions order to make debugging output more readable and cause-effect relationships visible; - making camperiphdone() (common device recovery completion handler) to also use cam_periph_error() in most cases, instead of own dumb code; - removing manual sense fetching code from cam_periph_error(); I was told by number of people that it is SIM obligation to fetch sense data, so this code is useless and only significantly complicates recovery logic; - making ada, da and pass driver to use cam_periph_error() with new limited recovery options to handle error recovery and debugging in common way; as one of results, CAM_REQUEUE_REQ and other retrying statuses are now working fine with pass driver, that caused many problems before. - reverting r186891 by raj@ to avoid burning few seconds in tight DELAY() loops on device probe, while device simply loads media; I think that problem may already be fixed in other way, and even if it is not, solution must be different. Sponsored by: iXsystems, Inc. MFC after: 2 weeks
2012-06-09 13:07:44 +00:00
return (error);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
Plumb device physical path reporting from CAM devices, through GEOM and DEVFS, and make it accessible via the diskinfo utility. Extend GEOM's generic attribute query mechanism into generic disk consumers. sys/geom/geom_disk.c: sys/geom/geom_disk.h: sys/cam/scsi/scsi_da.c: sys/cam/ata/ata_da.c: - Allow disk providers to implement a new method which can override the default BIO_GETATTR response, d_getattr(struct bio *). This function returns -1 if not handled, otherwise it returns 0 or an errno to be passed to g_io_deliver(). sys/cam/scsi/scsi_da.c: sys/cam/ata/ata_da.c: - Don't copy the serial number to dp->d_ident anymore, as the CAM XPT is now responsible for returning this information via d_getattr()->(a)dagetattr()->xpt_getatr(). sys/geom/geom_dev.c: - Implement a new ioctl, DIOCGPHYSPATH, which returns the GEOM attribute "GEOM::physpath", if possible. If the attribute request returns a zero-length string, ENOENT is returned. usr.sbin/diskinfo/diskinfo.c: - If the DIOCGPHYSPATH ioctl is successful, report physical path data when diskinfo is executed with the '-v' option. Submitted by: will Reviewed by: gibbs Sponsored by: Spectra Logic Corporation Add generic attribute change notification support to GEOM. sys/sys/geom/geom.h: Add a new attrchanged method field to both g_class and g_geom. sys/sys/geom/geom.h: sys/geom/geom_event.c: - Provide the g_attr_changed() function that providers can use to advertise attribute changes. - Perform delivery of attribute change notifications from a thread context via the standard GEOM event mechanism. sys/geom/geom_subr.c: Inherit the attrchanged method from class to geom (class instance). sys/geom/geom_disk.c: Provide disk_attr_changed() to provide g_attr_changed() access to consumers of the disk API. sys/cam/scsi/scsi_pass.c: sys/cam/scsi/scsi_da.c: sys/geom/geom_dev.c: sys/geom/geom_disk.c: Use attribute changed events to track updates to physical path information. sys/cam/scsi/scsi_da.c: Add AC_ADVINFO_CHANGED to the registered asynchronous CAM events for this driver. When this event occurs, and the updated buffer type references our physical path attribute, emit a GEOM attribute changed event via the disk_attr_changed() API. sys/cam/scsi/scsi_pass.c: Add AC_ADVINFO_CHANGED to the registered asynchronous CAM events for this driver. When this event occurs, update the physical patch devfs alias for this pass instance. Submitted by: gibbs Sponsored by: Spectra Logic Corporation
2011-06-14 17:10:32 +00:00
static int
dagetattr(struct bio *bp)
{
Protect xpt_getattr() calls with the SIM lock and assert that. Submitted by: ken@ (earlier version)
2012-10-12 17:18:24 +00:00
int ret;
Plumb device physical path reporting from CAM devices, through GEOM and DEVFS, and make it accessible via the diskinfo utility. Extend GEOM's generic attribute query mechanism into generic disk consumers. sys/geom/geom_disk.c: sys/geom/geom_disk.h: sys/cam/scsi/scsi_da.c: sys/cam/ata/ata_da.c: - Allow disk providers to implement a new method which can override the default BIO_GETATTR response, d_getattr(struct bio *). This function returns -1 if not handled, otherwise it returns 0 or an errno to be passed to g_io_deliver(). sys/cam/scsi/scsi_da.c: sys/cam/ata/ata_da.c: - Don't copy the serial number to dp->d_ident anymore, as the CAM XPT is now responsible for returning this information via d_getattr()->(a)dagetattr()->xpt_getatr(). sys/geom/geom_dev.c: - Implement a new ioctl, DIOCGPHYSPATH, which returns the GEOM attribute "GEOM::physpath", if possible. If the attribute request returns a zero-length string, ENOENT is returned. usr.sbin/diskinfo/diskinfo.c: - If the DIOCGPHYSPATH ioctl is successful, report physical path data when diskinfo is executed with the '-v' option. Submitted by: will Reviewed by: gibbs Sponsored by: Spectra Logic Corporation Add generic attribute change notification support to GEOM. sys/sys/geom/geom.h: Add a new attrchanged method field to both g_class and g_geom. sys/sys/geom/geom.h: sys/geom/geom_event.c: - Provide the g_attr_changed() function that providers can use to advertise attribute changes. - Perform delivery of attribute change notifications from a thread context via the standard GEOM event mechanism. sys/geom/geom_subr.c: Inherit the attrchanged method from class to geom (class instance). sys/geom/geom_disk.c: Provide disk_attr_changed() to provide g_attr_changed() access to consumers of the disk API. sys/cam/scsi/scsi_pass.c: sys/cam/scsi/scsi_da.c: sys/geom/geom_dev.c: sys/geom/geom_disk.c: Use attribute changed events to track updates to physical path information. sys/cam/scsi/scsi_da.c: Add AC_ADVINFO_CHANGED to the registered asynchronous CAM events for this driver. When this event occurs, and the updated buffer type references our physical path attribute, emit a GEOM attribute changed event via the disk_attr_changed() API. sys/cam/scsi/scsi_pass.c: Add AC_ADVINFO_CHANGED to the registered asynchronous CAM events for this driver. When this event occurs, update the physical patch devfs alias for this pass instance. Submitted by: gibbs Sponsored by: Spectra Logic Corporation
2011-06-14 17:10:32 +00:00
struct cam_periph *periph;
periph = (struct cam_periph *)bp->bio_disk->d_drv1;
Protect xpt_getattr() calls with the SIM lock and assert that. Submitted by: ken@ (earlier version)
2012-10-12 17:18:24 +00:00
cam_periph_lock(periph);
Plumb device physical path reporting from CAM devices, through GEOM and DEVFS, and make it accessible via the diskinfo utility. Extend GEOM's generic attribute query mechanism into generic disk consumers. sys/geom/geom_disk.c: sys/geom/geom_disk.h: sys/cam/scsi/scsi_da.c: sys/cam/ata/ata_da.c: - Allow disk providers to implement a new method which can override the default BIO_GETATTR response, d_getattr(struct bio *). This function returns -1 if not handled, otherwise it returns 0 or an errno to be passed to g_io_deliver(). sys/cam/scsi/scsi_da.c: sys/cam/ata/ata_da.c: - Don't copy the serial number to dp->d_ident anymore, as the CAM XPT is now responsible for returning this information via d_getattr()->(a)dagetattr()->xpt_getatr(). sys/geom/geom_dev.c: - Implement a new ioctl, DIOCGPHYSPATH, which returns the GEOM attribute "GEOM::physpath", if possible. If the attribute request returns a zero-length string, ENOENT is returned. usr.sbin/diskinfo/diskinfo.c: - If the DIOCGPHYSPATH ioctl is successful, report physical path data when diskinfo is executed with the '-v' option. Submitted by: will Reviewed by: gibbs Sponsored by: Spectra Logic Corporation Add generic attribute change notification support to GEOM. sys/sys/geom/geom.h: Add a new attrchanged method field to both g_class and g_geom. sys/sys/geom/geom.h: sys/geom/geom_event.c: - Provide the g_attr_changed() function that providers can use to advertise attribute changes. - Perform delivery of attribute change notifications from a thread context via the standard GEOM event mechanism. sys/geom/geom_subr.c: Inherit the attrchanged method from class to geom (class instance). sys/geom/geom_disk.c: Provide disk_attr_changed() to provide g_attr_changed() access to consumers of the disk API. sys/cam/scsi/scsi_pass.c: sys/cam/scsi/scsi_da.c: sys/geom/geom_dev.c: sys/geom/geom_disk.c: Use attribute changed events to track updates to physical path information. sys/cam/scsi/scsi_da.c: Add AC_ADVINFO_CHANGED to the registered asynchronous CAM events for this driver. When this event occurs, and the updated buffer type references our physical path attribute, emit a GEOM attribute changed event via the disk_attr_changed() API. sys/cam/scsi/scsi_pass.c: Add AC_ADVINFO_CHANGED to the registered asynchronous CAM events for this driver. When this event occurs, update the physical patch devfs alias for this pass instance. Submitted by: gibbs Sponsored by: Spectra Logic Corporation
2011-06-14 17:10:32 +00:00
ret = xpt_getattr(bp->bio_data, bp->bio_length, bp->bio_attribute,
periph->path);
Protect xpt_getattr() calls with the SIM lock and assert that. Submitted by: ken@ (earlier version)
2012-10-12 17:18:24 +00:00
cam_periph_unlock(periph);
Plumb device physical path reporting from CAM devices, through GEOM and DEVFS, and make it accessible via the diskinfo utility. Extend GEOM's generic attribute query mechanism into generic disk consumers. sys/geom/geom_disk.c: sys/geom/geom_disk.h: sys/cam/scsi/scsi_da.c: sys/cam/ata/ata_da.c: - Allow disk providers to implement a new method which can override the default BIO_GETATTR response, d_getattr(struct bio *). This function returns -1 if not handled, otherwise it returns 0 or an errno to be passed to g_io_deliver(). sys/cam/scsi/scsi_da.c: sys/cam/ata/ata_da.c: - Don't copy the serial number to dp->d_ident anymore, as the CAM XPT is now responsible for returning this information via d_getattr()->(a)dagetattr()->xpt_getatr(). sys/geom/geom_dev.c: - Implement a new ioctl, DIOCGPHYSPATH, which returns the GEOM attribute "GEOM::physpath", if possible. If the attribute request returns a zero-length string, ENOENT is returned. usr.sbin/diskinfo/diskinfo.c: - If the DIOCGPHYSPATH ioctl is successful, report physical path data when diskinfo is executed with the '-v' option. Submitted by: will Reviewed by: gibbs Sponsored by: Spectra Logic Corporation Add generic attribute change notification support to GEOM. sys/sys/geom/geom.h: Add a new attrchanged method field to both g_class and g_geom. sys/sys/geom/geom.h: sys/geom/geom_event.c: - Provide the g_attr_changed() function that providers can use to advertise attribute changes. - Perform delivery of attribute change notifications from a thread context via the standard GEOM event mechanism. sys/geom/geom_subr.c: Inherit the attrchanged method from class to geom (class instance). sys/geom/geom_disk.c: Provide disk_attr_changed() to provide g_attr_changed() access to consumers of the disk API. sys/cam/scsi/scsi_pass.c: sys/cam/scsi/scsi_da.c: sys/geom/geom_dev.c: sys/geom/geom_disk.c: Use attribute changed events to track updates to physical path information. sys/cam/scsi/scsi_da.c: Add AC_ADVINFO_CHANGED to the registered asynchronous CAM events for this driver. When this event occurs, and the updated buffer type references our physical path attribute, emit a GEOM attribute changed event via the disk_attr_changed() API. sys/cam/scsi/scsi_pass.c: Add AC_ADVINFO_CHANGED to the registered asynchronous CAM events for this driver. When this event occurs, update the physical patch devfs alias for this pass instance. Submitted by: gibbs Sponsored by: Spectra Logic Corporation
2011-06-14 17:10:32 +00:00
if (ret == 0)
bp->bio_completed = bp->bio_length;
return ret;
}
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
static void
dainit(void)
{
cam_status status;
/*
* Install a global async callback. This callback will
* receive async callbacks like "new device found".
*/
Add a helper function for registering async callbacks. Besides eliminating a lot of duplicated code, this also fixes a locking edge case.
2007-05-16 16:54:23 +00:00
status = xpt_register_async(AC_FOUND_DEVICE, daasync, NULL, NULL);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
if (status != CAM_REQ_CMP) {
printf("da: Failed to attach master async callback "
"due to status 0x%x!\n", status);
Add a tunable that allows one to turn off the automatic sending of the ORDERED tag. This recoups significant performance gains for many arrays. The default is still to send out the ORDERED tag periodically. Reviewed by: scsi (justin+timeout)
2006-11-02 21:12:37 +00:00
} else if (da_send_ordered) {
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
Implement a new generic mechanism for attaching handler functions to events, in order to pave the way for removing a number of the ad-hoc implementations currently in use. Retire the at_shutdown family of functions and replace them with new event handler lists. Rework kern_shutdown.c to take greater advantage of the use of event handlers. Reviewed by: green
1999-08-21 06:24:40 +00:00
/* Register our shutdown event handler */
if ((EVENTHANDLER_REGISTER(shutdown_post_sync, dashutdown,
Back out last commit.
2003-01-06 19:30:21 +00:00
NULL, SHUTDOWN_PRI_DEFAULT)) == NULL)
Implement a new generic mechanism for attaching handler functions to events, in order to pave the way for removing a number of the ad-hoc implementations currently in use. Retire the at_shutdown family of functions and replace them with new event handler lists. Rework kern_shutdown.c to take greater advantage of the use of event handlers. Reviewed by: green
1999-08-21 06:24:40 +00:00
printf("dainit: shutdown event registration failed!\n");
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
}
Fix a bug which causes a panic in daopen(). The panic is caused by a da(4) instance going away while GEOM is still probing it. In this case, the GEOM disk class instance has been created by disk_create(), and the taste of the disk is queued in the GEOM event queue. While that event is queued, the da(4) instance goes away. When the open call comes into the da(4) driver, it dereferences the freed (but non-NULL) peripheral pointer provided by GEOM, which results in a panic. The solution is to add a callback to the GEOM disk code that is called when all of its resources are cleaned up. This is implemented inside GEOM by adding an optional callback that is called when all consumers have detached from a provider, and the provider is about to be deleted. scsi_cd.c, scsi_da.c: In the register routine for the cd(4) and da(4) routines, acquire a reference to the CAM peripheral instance just before we call disk_create(). Use the new GEOM disk d_gone() callback to register a callback (dadiskgonecb()/cddiskgonecb()) that decrements the peripheral reference count once GEOM has finished cleaning up its resources. In the cd(4) driver, clean up open and close behavior slightly. GEOM makes sure we only get one open() and one close call, so there is no need to set an open flag and decrement the reference count if we are not the first open. In the cd(4) driver, use cam_periph_release_locked() in a couple of error scenarios to avoid extra mutex calls. geom.h: Add a new, optional, providergone callback that is called when a provider is about to be deleted. geom_disk.h: Add a new d_gone() callback to the GEOM disk interface. Bump the DISK_VERSION to version 2. This probably should have been done after a couple of previous changes, especially the addition of the d_getattr() callback. geom_disk.c: Add a providergone callback for the disk class, g_disk_providergone(), that calls the user's d_gone() callback if it exists. Bump the DISK_VERSION to 2. geom_subr.c: In g_destroy_provider(), call the providergone callback if it has been provided. In g_new_geomf(), propagate the class's providergone callback to the new geom instance. blkfront.c: Callers of disk_create() are supposed to pass in DISK_VERSION, not an explicit disk API version number. Update the blkfront driver to do that. disk.9: Update the disk(9) man page to include information on the new d_gone() callback, as well as the previously added d_getattr() callback, d_descr field, and HBA PCI ID fields. MFC after: 5 days
2012-06-24 04:29:03 +00:00
/*
* Callback from GEOM, called when it has finished cleaning up its
* resources.
*/
static void
dadiskgonecb(struct disk *dp)
{
struct cam_periph *periph;
periph = (struct cam_periph *)dp->d_drv1;
cam_periph_release(periph);
}
Fix a problem with the way we handled device invalidation when attaching to a device failed. In theory, the same steps that happen when we get an AC_LOST_DEVICE async notification should have been taken when a driver fails to attach. In practice, that wasn't the case. This only affected the da, cd and ch drivers, but the fix affects all peripheral drivers. There were several possible problems: - In the da driver, we didn't remove the peripheral's softc from the da driver's linked list of softcs. Once the peripheral and softc got removed, we'd get a kernel panic the next time the timeout routine called dasendorderedtag(). - In the da, cd and possibly ch drivers, we didn't remove the peripheral's devstat structure from the devstat queue. Once the peripheral and softc were removed, this could cause a panic if anyone tried to access device statistics. (one component of the linked list wouldn't exist anymore) - In the cd driver, we didn't take the peripheral off the changer run queue if it was scheduled to run. In practice, it's highly unlikely, and maybe impossible that the peripheral would have been on the changer run queue at that stage of the probe process. The fix is: - Add a new peripheral callback function (the "oninvalidate" function) that is called the first time cam_periph_invalidate() is called for a peripheral. - Create new foooninvalidate() routines for each peripheral driver. This routine is always called at splsoftcam(), and contains all the stuff that used to be in the AC_LOST_DEVICE case of the async callback handler. - Move the devstat cleanup call to the destructor/cleanup routines, since some of the drivers do I/O in their close routines. - Make sure that when we're flushing the buffer queue, we traverse it at splbio(). - Add a check for the invalid flag in the pt driver's open routine. Reviewed by: gibbs
1998-10-22 22:16:56 +00:00
static void
daoninvalidate(struct cam_periph *periph)
{
struct da_softc *softc;
softc = (struct da_softc *)periph->softc;
/*
* De-register any async callbacks.
*/
Add a helper function for registering async callbacks. Besides eliminating a lot of duplicated code, this also fixes a locking edge case.
2007-05-16 16:54:23 +00:00
xpt_register_async(0, daasync, periph, periph->path);
Fix a problem with the way we handled device invalidation when attaching to a device failed. In theory, the same steps that happen when we get an AC_LOST_DEVICE async notification should have been taken when a driver fails to attach. In practice, that wasn't the case. This only affected the da, cd and ch drivers, but the fix affects all peripheral drivers. There were several possible problems: - In the da driver, we didn't remove the peripheral's softc from the da driver's linked list of softcs. Once the peripheral and softc got removed, we'd get a kernel panic the next time the timeout routine called dasendorderedtag(). - In the da, cd and possibly ch drivers, we didn't remove the peripheral's devstat structure from the devstat queue. Once the peripheral and softc were removed, this could cause a panic if anyone tried to access device statistics. (one component of the linked list wouldn't exist anymore) - In the cd driver, we didn't take the peripheral off the changer run queue if it was scheduled to run. In practice, it's highly unlikely, and maybe impossible that the peripheral would have been on the changer run queue at that stage of the probe process. The fix is: - Add a new peripheral callback function (the "oninvalidate" function) that is called the first time cam_periph_invalidate() is called for a peripheral. - Create new foooninvalidate() routines for each peripheral driver. This routine is always called at splsoftcam(), and contains all the stuff that used to be in the AC_LOST_DEVICE case of the async callback handler. - Move the devstat cleanup call to the destructor/cleanup routines, since some of the drivers do I/O in their close routines. - Make sure that when we're flushing the buffer queue, we traverse it at splbio(). - Add a check for the invalid flag in the pt driver's open routine. Reviewed by: gibbs
1998-10-22 22:16:56 +00:00
softc->flags |= DA_FLAG_PACK_INVALID;
/*
* Return all queued I/O with ENXIO.
Revert r211434. Offline discussions have convinced me that this should be left alone for now.
2010-08-20 17:20:05 +00:00
* XXX Handle any transactions queued to the card
* with XPT_ABORT_CCB.
Fix a problem with the way we handled device invalidation when attaching to a device failed. In theory, the same steps that happen when we get an AC_LOST_DEVICE async notification should have been taken when a driver fails to attach. In practice, that wasn't the case. This only affected the da, cd and ch drivers, but the fix affects all peripheral drivers. There were several possible problems: - In the da driver, we didn't remove the peripheral's softc from the da driver's linked list of softcs. Once the peripheral and softc got removed, we'd get a kernel panic the next time the timeout routine called dasendorderedtag(). - In the da, cd and possibly ch drivers, we didn't remove the peripheral's devstat structure from the devstat queue. Once the peripheral and softc were removed, this could cause a panic if anyone tried to access device statistics. (one component of the linked list wouldn't exist anymore) - In the cd driver, we didn't take the peripheral off the changer run queue if it was scheduled to run. In practice, it's highly unlikely, and maybe impossible that the peripheral would have been on the changer run queue at that stage of the probe process. The fix is: - Add a new peripheral callback function (the "oninvalidate" function) that is called the first time cam_periph_invalidate() is called for a peripheral. - Create new foooninvalidate() routines for each peripheral driver. This routine is always called at splsoftcam(), and contains all the stuff that used to be in the AC_LOST_DEVICE case of the async callback handler. - Move the devstat cleanup call to the destructor/cleanup routines, since some of the drivers do I/O in their close routines. - Make sure that when we're flushing the buffer queue, we traverse it at splbio(). - Add a check for the invalid flag in the pt driver's open routine. Reviewed by: gibbs
1998-10-22 22:16:56 +00:00
*/
Use bioq_flush() to drain a bio queue with a specific error code. Retain the mistake of not updating the devstat API for now. Spell bioq_disksort() consistently with the remaining bioq_*(). #include <geom/geom_disk.h> where this is more appropriate.
2003-04-01 15:06:26 +00:00
bioq_flush(&softc->bio_queue, NULL, ENXIO);
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
bioq_flush(&softc->delete_queue, NULL, ENXIO);
Fix a problem with the way we handled device invalidation when attaching to a device failed. In theory, the same steps that happen when we get an AC_LOST_DEVICE async notification should have been taken when a driver fails to attach. In practice, that wasn't the case. This only affected the da, cd and ch drivers, but the fix affects all peripheral drivers. There were several possible problems: - In the da driver, we didn't remove the peripheral's softc from the da driver's linked list of softcs. Once the peripheral and softc got removed, we'd get a kernel panic the next time the timeout routine called dasendorderedtag(). - In the da, cd and possibly ch drivers, we didn't remove the peripheral's devstat structure from the devstat queue. Once the peripheral and softc were removed, this could cause a panic if anyone tried to access device statistics. (one component of the linked list wouldn't exist anymore) - In the cd driver, we didn't take the peripheral off the changer run queue if it was scheduled to run. In practice, it's highly unlikely, and maybe impossible that the peripheral would have been on the changer run queue at that stage of the probe process. The fix is: - Add a new peripheral callback function (the "oninvalidate" function) that is called the first time cam_periph_invalidate() is called for a peripheral. - Create new foooninvalidate() routines for each peripheral driver. This routine is always called at splsoftcam(), and contains all the stuff that used to be in the AC_LOST_DEVICE case of the async callback handler. - Move the devstat cleanup call to the destructor/cleanup routines, since some of the drivers do I/O in their close routines. - Make sure that when we're flushing the buffer queue, we traverse it at splbio(). - Add a check for the invalid flag in the pt driver's open routine. Reviewed by: gibbs
1998-10-22 22:16:56 +00:00
Fix a bug which causes a panic in daopen(). The panic is caused by a da(4) instance going away while GEOM is still probing it. In this case, the GEOM disk class instance has been created by disk_create(), and the taste of the disk is queued in the GEOM event queue. While that event is queued, the da(4) instance goes away. When the open call comes into the da(4) driver, it dereferences the freed (but non-NULL) peripheral pointer provided by GEOM, which results in a panic. The solution is to add a callback to the GEOM disk code that is called when all of its resources are cleaned up. This is implemented inside GEOM by adding an optional callback that is called when all consumers have detached from a provider, and the provider is about to be deleted. scsi_cd.c, scsi_da.c: In the register routine for the cd(4) and da(4) routines, acquire a reference to the CAM peripheral instance just before we call disk_create(). Use the new GEOM disk d_gone() callback to register a callback (dadiskgonecb()/cddiskgonecb()) that decrements the peripheral reference count once GEOM has finished cleaning up its resources. In the cd(4) driver, clean up open and close behavior slightly. GEOM makes sure we only get one open() and one close call, so there is no need to set an open flag and decrement the reference count if we are not the first open. In the cd(4) driver, use cam_periph_release_locked() in a couple of error scenarios to avoid extra mutex calls. geom.h: Add a new, optional, providergone callback that is called when a provider is about to be deleted. geom_disk.h: Add a new d_gone() callback to the GEOM disk interface. Bump the DISK_VERSION to version 2. This probably should have been done after a couple of previous changes, especially the addition of the d_getattr() callback. geom_disk.c: Add a providergone callback for the disk class, g_disk_providergone(), that calls the user's d_gone() callback if it exists. Bump the DISK_VERSION to 2. geom_subr.c: In g_destroy_provider(), call the providergone callback if it has been provided. In g_new_geomf(), propagate the class's providergone callback to the new geom instance. blkfront.c: Callers of disk_create() are supposed to pass in DISK_VERSION, not an explicit disk API version number. Update the blkfront driver to do that. disk.9: Update the disk(9) man page to include information on the new d_gone() callback, as well as the previously added d_getattr() callback, d_descr field, and HBA PCI ID fields. MFC after: 5 days
2012-06-24 04:29:03 +00:00
/*
* Tell GEOM that we've gone away, we'll get a callback when it is
* done cleaning up its resources.
*/
Fix a bug that caused some /dev entries to continue to exist after the underlying drive had been hot-unplugged from the system. Here is a specific example. Filesystem code had opened /dev/da1s1e. Subsequently, the drive was hot-unplugged. This (correctly) caused all of the associated /dev/da1* entries to be deleted. When the filesystem later realized that the drive was gone it closed the device, reducing the write-access counts to 0 on the geom providers for da1s1e, da1s1, and da1. This caused geom to re-taste the providers, resulting in the devices being created again. When the drive was hot-plugged back in, it resulted in duplicate /dev entries for da1s1e, da1s1, and da1. This fix adds a new disk_gone() function which is called by CAM when a drive goes away. It orphans all of the providers associated with the drive, setting an error condition of ENXIO in each one. In addition, we prevent a re-taste on last close for writing if an error condition has been set in the provider. Sponsored by: Isilon Systems Reviewed by: phk MFC after: 1 week
2005-11-18 02:43:49 +00:00
disk_gone(softc->disk);
Fix a problem with the way we handled device invalidation when attaching to a device failed. In theory, the same steps that happen when we get an AC_LOST_DEVICE async notification should have been taken when a driver fails to attach. In practice, that wasn't the case. This only affected the da, cd and ch drivers, but the fix affects all peripheral drivers. There were several possible problems: - In the da driver, we didn't remove the peripheral's softc from the da driver's linked list of softcs. Once the peripheral and softc got removed, we'd get a kernel panic the next time the timeout routine called dasendorderedtag(). - In the da, cd and possibly ch drivers, we didn't remove the peripheral's devstat structure from the devstat queue. Once the peripheral and softc were removed, this could cause a panic if anyone tried to access device statistics. (one component of the linked list wouldn't exist anymore) - In the cd driver, we didn't take the peripheral off the changer run queue if it was scheduled to run. In practice, it's highly unlikely, and maybe impossible that the peripheral would have been on the changer run queue at that stage of the probe process. The fix is: - Add a new peripheral callback function (the "oninvalidate" function) that is called the first time cam_periph_invalidate() is called for a peripheral. - Create new foooninvalidate() routines for each peripheral driver. This routine is always called at splsoftcam(), and contains all the stuff that used to be in the AC_LOST_DEVICE case of the async callback handler. - Move the devstat cleanup call to the destructor/cleanup routines, since some of the drivers do I/O in their close routines. - Make sure that when we're flushing the buffer queue, we traverse it at splbio(). - Add a check for the invalid flag in the pt driver's open routine. Reviewed by: gibbs
1998-10-22 22:16:56 +00:00
}
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
static void
dacleanup(struct cam_periph *periph)
{
Fix a problem with the way we handled device invalidation when attaching to a device failed. In theory, the same steps that happen when we get an AC_LOST_DEVICE async notification should have been taken when a driver fails to attach. In practice, that wasn't the case. This only affected the da, cd and ch drivers, but the fix affects all peripheral drivers. There were several possible problems: - In the da driver, we didn't remove the peripheral's softc from the da driver's linked list of softcs. Once the peripheral and softc got removed, we'd get a kernel panic the next time the timeout routine called dasendorderedtag(). - In the da, cd and possibly ch drivers, we didn't remove the peripheral's devstat structure from the devstat queue. Once the peripheral and softc were removed, this could cause a panic if anyone tried to access device statistics. (one component of the linked list wouldn't exist anymore) - In the cd driver, we didn't take the peripheral off the changer run queue if it was scheduled to run. In practice, it's highly unlikely, and maybe impossible that the peripheral would have been on the changer run queue at that stage of the probe process. The fix is: - Add a new peripheral callback function (the "oninvalidate" function) that is called the first time cam_periph_invalidate() is called for a peripheral. - Create new foooninvalidate() routines for each peripheral driver. This routine is always called at splsoftcam(), and contains all the stuff that used to be in the AC_LOST_DEVICE case of the async callback handler. - Move the devstat cleanup call to the destructor/cleanup routines, since some of the drivers do I/O in their close routines. - Make sure that when we're flushing the buffer queue, we traverse it at splbio(). - Add a check for the invalid flag in the pt driver's open routine. Reviewed by: gibbs
1998-10-22 22:16:56 +00:00
struct da_softc *softc;
softc = (struct da_softc *)periph->softc;
Reorder dacleanup() and cdcleanup() slightly so that the sysctl context is freed while the periph lock is not held. While here, wait until after freeing the softc before reacquiring the periph lock. Tested by: sbruno
2009-02-11 22:29:09 +00:00
cam_periph_unlock(periph);
Clean up dynamically allocated sysctl variables when we run dacleanup() and cdcleanup(). This fixes sysctl problems ("can't re-use a leaf") when someone adds another peripheral at the same unit number. (e.g. rescan da0, it goes away, then rescan again and da0 comes back, but since we haven't cleaned up the sysctl variables from the last da0 instance, we can't register the variables for the new instance under the same name.) Reported by: njl Tested by: njl
2003-03-26 04:38:39 +00:00
/*
* If we can't free the sysctl tree, oh well...
*/
(dacleanup, dasysctlinit): Defend against calling sysctl_ctx_free on an uninitialized sysctl_ctx, using flag DA_FLAG_SCTX_INIT. This prevents a panic encoutered with some umass units that probe correctly but fail to attach. Same problem, and same fix, as scsi_cd.c rev. 1.86. Reviewed by: njl, ken
2003-10-08 07:12:30 +00:00
if ((softc->flags & DA_FLAG_SCTX_INIT) != 0
&& sysctl_ctx_free(&softc->sysctl_ctx) != 0) {
Add an xpt_print function to reduce most of the xpt_print_path/printf pairs. Convert the core code to use it. Reviewed by: scsi (various folks weighing in with good advice) MFC after: 2 weeks
2006-12-05 07:45:28 +00:00
xpt_print(periph->path, "can't remove sysctl context\n");
Clean up dynamically allocated sysctl variables when we run dacleanup() and cdcleanup(). This fixes sysctl problems ("can't re-use a leaf") when someone adds another peripheral at the same unit number. (e.g. rescan da0, it goes away, then rescan again and da0 comes back, but since we haven't cleaned up the sysctl variables from the last da0 instance, we can't register the variables for the new instance under the same name.) Reported by: njl Tested by: njl
2003-03-26 04:38:39 +00:00
}
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
callout_drain(&softc->mediapoll_c);
Drop the periph/sim lock when calling disk_destroy().
2007-04-16 19:41:14 +00:00
disk_destroy(softc->disk);
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
callout_drain(&softc->sendordered_c);
Fix a problem with the way we handled device invalidation when attaching to a device failed. In theory, the same steps that happen when we get an AC_LOST_DEVICE async notification should have been taken when a driver fails to attach. In practice, that wasn't the case. This only affected the da, cd and ch drivers, but the fix affects all peripheral drivers. There were several possible problems: - In the da driver, we didn't remove the peripheral's softc from the da driver's linked list of softcs. Once the peripheral and softc got removed, we'd get a kernel panic the next time the timeout routine called dasendorderedtag(). - In the da, cd and possibly ch drivers, we didn't remove the peripheral's devstat structure from the devstat queue. Once the peripheral and softc were removed, this could cause a panic if anyone tried to access device statistics. (one component of the linked list wouldn't exist anymore) - In the cd driver, we didn't take the peripheral off the changer run queue if it was scheduled to run. In practice, it's highly unlikely, and maybe impossible that the peripheral would have been on the changer run queue at that stage of the probe process. The fix is: - Add a new peripheral callback function (the "oninvalidate" function) that is called the first time cam_periph_invalidate() is called for a peripheral. - Create new foooninvalidate() routines for each peripheral driver. This routine is always called at splsoftcam(), and contains all the stuff that used to be in the AC_LOST_DEVICE case of the async callback handler. - Move the devstat cleanup call to the destructor/cleanup routines, since some of the drivers do I/O in their close routines. - Make sure that when we're flushing the buffer queue, we traverse it at splbio(). - Add a check for the invalid flag in the pt driver's open routine. Reviewed by: gibbs
1998-10-22 22:16:56 +00:00
free(softc, M_DEVBUF);
Reorder dacleanup() and cdcleanup() slightly so that the sysctl context is freed while the periph lock is not held. While here, wait until after freeing the softc before reacquiring the periph lock. Tested by: sbruno
2009-02-11 22:29:09 +00:00
cam_periph_lock(periph);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
static void
daasync(void *callback_arg, u_int32_t code,
struct cam_path *path, void *arg)
{
struct cam_periph *periph;
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
struct da_softc *softc;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
periph = (struct cam_periph *)callback_arg;
switch (code) {
case AC_FOUND_DEVICE:
{
struct ccb_getdev *cgd;
cam_status status;
cgd = (struct ccb_getdev *)arg;
Check the void * argument in the AC_FOUND_DEV case against NULL. Whether correctly or not, this sometimes is propagated up via XPT.
2001-07-04 05:22:42 +00:00
if (cgd == NULL)
break;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
Separate the parallel scsi knowledge out of the core of the XPT, and modularize it so that new transports can be created. Add a transport for SATA Add a periph+protocol layer for ATA Add a driver for AHCI-compliant hardware. Add a maxio field to CAM so that drivers can advertise their max I/O capability. Modify various drivers so that they are insulated from the value of MAXPHYS. The new ATA/SATA code supports AHCI-compliant hardware, and will override the classic ATA driver if it is loaded as a module at boot time or compiled into the kernel. The stack now support NCQ (tagged queueing) for increased performance on modern SATA drives. It also supports port multipliers. ATA drives are accessed via 'ada' device nodes. ATAPI drives are accessed via 'cd' device nodes. They can all be enumerated and manipulated via camcontrol, just like SCSI drives. SCSI commands are not translated to their ATA equivalents; ATA native commands are used throughout the entire stack, including camcontrol. See the camcontrol manpage for further details. Testing this code may require that you update your fstab, and possibly modify your BIOS to enable AHCI functionality, if available. This code is very experimental at the moment. The userland ABI/API has changed, so applications will need to be recompiled. It may change further in the near future. The 'ada' device name may also change as more infrastructure is completed in this project. The goal is to eventually put all CAM busses and devices until newbus, allowing for interesting topology and management options. Few functional changes will be seen with existing SCSI/SAS/FC drivers, though the userland ABI has still changed. In the future, transports specific modules for SAS and FC may appear in order to better support the topologies and capabilities of these technologies. The modularization of CAM and the addition of the ATA/SATA modules is meant to break CAM out of the mold of being specific to SCSI, letting it grow to be a framework for arbitrary transports and protocols. It also allows drivers to be written to support discrete hardware without jeopardizing the stability of non-related hardware. While only an AHCI driver is provided now, a Silicon Image driver is also in the works. Drivers for ICH1-4, ICH5-6, PIIX, classic IDE, and any other hardware is possible and encouraged. Help with new transports is also encouraged. Submitted by: scottl, mav Approved by: re
2009-07-10 08:18:08 +00:00
if (cgd->protocol != PROTO_SCSI)
break;
Attach pass driver to LUNs is OFFLINE state. Previously such LUNs were silently ignored. But while they indeed unable to process most of SCSI commands, some, like RTPG, they still can. MFC after: 1 month
2015-08-29 11:21:20 +00:00
if (SID_QUAL(&cgd->inq_data) != SID_QUAL_LU_CONNECTED)
break;
Do the minor changes needed because of change to ccb_getdev structure. JKH Trading Stamps applied. Reviewed by: gibbs@freebsd.org, ken@freebsd.org
2000-01-17 06:27:37 +00:00
if (SID_TYPE(&cgd->inq_data) != T_DIRECT
- Add support for Simplified Direct Access Device, mostly for Firewire/SBP-II devices. - Add quirk for Logitec USB/Firewire HDD. MFC after: 3 days.
2002-02-21 11:58:47 +00:00
&& SID_TYPE(&cgd->inq_data) != T_RBC
Do the minor changes needed because of change to ccb_getdev structure. JKH Trading Stamps applied. Reviewed by: gibbs@freebsd.org, ken@freebsd.org
2000-01-17 06:27:37 +00:00
&& SID_TYPE(&cgd->inq_data) != T_OPTICAL)
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
break;
/*
* Allocate a peripheral instance for
* this device and start the probe
* process.
*/
Fix a problem with the way we handled device invalidation when attaching to a device failed. In theory, the same steps that happen when we get an AC_LOST_DEVICE async notification should have been taken when a driver fails to attach. In practice, that wasn't the case. This only affected the da, cd and ch drivers, but the fix affects all peripheral drivers. There were several possible problems: - In the da driver, we didn't remove the peripheral's softc from the da driver's linked list of softcs. Once the peripheral and softc got removed, we'd get a kernel panic the next time the timeout routine called dasendorderedtag(). - In the da, cd and possibly ch drivers, we didn't remove the peripheral's devstat structure from the devstat queue. Once the peripheral and softc were removed, this could cause a panic if anyone tried to access device statistics. (one component of the linked list wouldn't exist anymore) - In the cd driver, we didn't take the peripheral off the changer run queue if it was scheduled to run. In practice, it's highly unlikely, and maybe impossible that the peripheral would have been on the changer run queue at that stage of the probe process. The fix is: - Add a new peripheral callback function (the "oninvalidate" function) that is called the first time cam_periph_invalidate() is called for a peripheral. - Create new foooninvalidate() routines for each peripheral driver. This routine is always called at splsoftcam(), and contains all the stuff that used to be in the AC_LOST_DEVICE case of the async callback handler. - Move the devstat cleanup call to the destructor/cleanup routines, since some of the drivers do I/O in their close routines. - Make sure that when we're flushing the buffer queue, we traverse it at splbio(). - Add a check for the invalid flag in the pt driver's open routine. Reviewed by: gibbs
1998-10-22 22:16:56 +00:00
status = cam_periph_alloc(daregister, daoninvalidate,
dacleanup, dastart,
"da", CAM_PERIPH_BIO,
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
path, daasync,
Fix a problem with the way we handled device invalidation when attaching to a device failed. In theory, the same steps that happen when we get an AC_LOST_DEVICE async notification should have been taken when a driver fails to attach. In practice, that wasn't the case. This only affected the da, cd and ch drivers, but the fix affects all peripheral drivers. There were several possible problems: - In the da driver, we didn't remove the peripheral's softc from the da driver's linked list of softcs. Once the peripheral and softc got removed, we'd get a kernel panic the next time the timeout routine called dasendorderedtag(). - In the da, cd and possibly ch drivers, we didn't remove the peripheral's devstat structure from the devstat queue. Once the peripheral and softc were removed, this could cause a panic if anyone tried to access device statistics. (one component of the linked list wouldn't exist anymore) - In the cd driver, we didn't take the peripheral off the changer run queue if it was scheduled to run. In practice, it's highly unlikely, and maybe impossible that the peripheral would have been on the changer run queue at that stage of the probe process. The fix is: - Add a new peripheral callback function (the "oninvalidate" function) that is called the first time cam_periph_invalidate() is called for a peripheral. - Create new foooninvalidate() routines for each peripheral driver. This routine is always called at splsoftcam(), and contains all the stuff that used to be in the AC_LOST_DEVICE case of the async callback handler. - Move the devstat cleanup call to the destructor/cleanup routines, since some of the drivers do I/O in their close routines. - Make sure that when we're flushing the buffer queue, we traverse it at splbio(). - Add a check for the invalid flag in the pt driver's open routine. Reviewed by: gibbs
1998-10-22 22:16:56 +00:00
AC_FOUND_DEVICE, cgd);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
if (status != CAM_REQ_CMP
&& status != CAM_REQ_INPROG)
printf("daasync: Unable to attach to new device "
"due to status 0x%x\n", status);
Plumb device physical path reporting from CAM devices, through GEOM and DEVFS, and make it accessible via the diskinfo utility. Extend GEOM's generic attribute query mechanism into generic disk consumers. sys/geom/geom_disk.c: sys/geom/geom_disk.h: sys/cam/scsi/scsi_da.c: sys/cam/ata/ata_da.c: - Allow disk providers to implement a new method which can override the default BIO_GETATTR response, d_getattr(struct bio *). This function returns -1 if not handled, otherwise it returns 0 or an errno to be passed to g_io_deliver(). sys/cam/scsi/scsi_da.c: sys/cam/ata/ata_da.c: - Don't copy the serial number to dp->d_ident anymore, as the CAM XPT is now responsible for returning this information via d_getattr()->(a)dagetattr()->xpt_getatr(). sys/geom/geom_dev.c: - Implement a new ioctl, DIOCGPHYSPATH, which returns the GEOM attribute "GEOM::physpath", if possible. If the attribute request returns a zero-length string, ENOENT is returned. usr.sbin/diskinfo/diskinfo.c: - If the DIOCGPHYSPATH ioctl is successful, report physical path data when diskinfo is executed with the '-v' option. Submitted by: will Reviewed by: gibbs Sponsored by: Spectra Logic Corporation Add generic attribute change notification support to GEOM. sys/sys/geom/geom.h: Add a new attrchanged method field to both g_class and g_geom. sys/sys/geom/geom.h: sys/geom/geom_event.c: - Provide the g_attr_changed() function that providers can use to advertise attribute changes. - Perform delivery of attribute change notifications from a thread context via the standard GEOM event mechanism. sys/geom/geom_subr.c: Inherit the attrchanged method from class to geom (class instance). sys/geom/geom_disk.c: Provide disk_attr_changed() to provide g_attr_changed() access to consumers of the disk API. sys/cam/scsi/scsi_pass.c: sys/cam/scsi/scsi_da.c: sys/geom/geom_dev.c: sys/geom/geom_disk.c: Use attribute changed events to track updates to physical path information. sys/cam/scsi/scsi_da.c: Add AC_ADVINFO_CHANGED to the registered asynchronous CAM events for this driver. When this event occurs, and the updated buffer type references our physical path attribute, emit a GEOM attribute changed event via the disk_attr_changed() API. sys/cam/scsi/scsi_pass.c: Add AC_ADVINFO_CHANGED to the registered asynchronous CAM events for this driver. When this event occurs, update the physical patch devfs alias for this pass instance. Submitted by: gibbs Sponsored by: Spectra Logic Corporation
2011-06-14 17:10:32 +00:00
return;
}
case AC_ADVINFO_CHANGED:
{
uintptr_t buftype;
buftype = (uintptr_t)arg;
if (buftype == CDAI_TYPE_PHYS_PATH) {
struct da_softc *softc;
softc = periph->softc;
disk_attr_changed(softc->disk, "GEOM::physpath",
M_NOWAIT);
}
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
break;
}
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
case AC_UNIT_ATTENTION:
{
union ccb *ccb;
int error_code, sense_key, asc, ascq;
softc = (struct da_softc *)periph->softc;
ccb = (union ccb *)arg;
/*
* Handle all UNIT ATTENTIONs except our own,
* as they will be handled by daerror().
*/
if (xpt_path_periph(ccb->ccb_h.path) != periph &&
scsi_extract_sense_ccb(ccb,
&error_code, &sense_key, &asc, &ascq)) {
if (asc == 0x2A && ascq == 0x09) {
xpt_print(ccb->ccb_h.path,
Do not reread SCSI disk VPD pages on every device open. Instead of rereading VPD pages on every device open, do it only on initial device probe, and in cases when device reported via UNIT ATTENTIONs that something has changed. Capacity is still rereaded on every open because it is more critical for operation and more probable to change in run time. On my tests with Intel 530 SSDs on mps(4) HBA this change reduces time GEOM needs to retaste the device (that includes few open/close cycles) from ~150ms to ~30ms. MFC after: 2 weeks Sponsored by: iXsystems, Inc.
2014-04-30 17:38:26 +00:00
"Capacity data has changed\n");
softc->flags &= ~DA_FLAG_PROBED;
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
dareprobe(periph);
Do not reread SCSI disk VPD pages on every device open. Instead of rereading VPD pages on every device open, do it only on initial device probe, and in cases when device reported via UNIT ATTENTIONs that something has changed. Capacity is still rereaded on every open because it is more critical for operation and more probable to change in run time. On my tests with Intel 530 SSDs on mps(4) HBA this change reduces time GEOM needs to retaste the device (that includes few open/close cycles) from ~150ms to ~30ms. MFC after: 2 weeks Sponsored by: iXsystems, Inc.
2014-04-30 17:38:26 +00:00
} else if (asc == 0x28 && ascq == 0x00) {
softc->flags &= ~DA_FLAG_PROBED;
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
disk_media_changed(softc->disk, M_NOWAIT);
Do not reread SCSI disk VPD pages on every device open. Instead of rereading VPD pages on every device open, do it only on initial device probe, and in cases when device reported via UNIT ATTENTIONs that something has changed. Capacity is still rereaded on every open because it is more critical for operation and more probable to change in run time. On my tests with Intel 530 SSDs on mps(4) HBA this change reduces time GEOM needs to retaste the device (that includes few open/close cycles) from ~150ms to ~30ms. MFC after: 2 weeks Sponsored by: iXsystems, Inc.
2014-04-30 17:38:26 +00:00
} else if (asc == 0x3F && ascq == 0x03) {
xpt_print(ccb->ccb_h.path,
"INQUIRY data has changed\n");
softc->flags &= ~DA_FLAG_PROBED;
dareprobe(periph);
}
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
}
cam_periph_async(periph, code, path, arg);
break;
}
case AC_SCSI_AEN:
softc = (struct da_softc *)periph->softc;
Do not schedule periph for payload/TUR requests if reprobe is in progress to avoid sending extra READ CAPACITY requests by dastart(). Schedule periph again on reprobe completion, or otherwise it may stuck indefinitely long. This should fix USB explore thread hanging on device unplug, waiting for periph destruction. Reported by: hselasky
2013-01-11 16:10:11 +00:00
if (!softc->tur) {
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
if (cam_periph_acquire(periph) == CAM_REQ_CMP) {
softc->tur = 1;
Do not schedule periph for payload/TUR requests if reprobe is in progress to avoid sending extra READ CAPACITY requests by dastart(). Schedule periph again on reprobe completion, or otherwise it may stuck indefinitely long. This should fix USB explore thread hanging on device unplug, waiting for periph destruction. Reported by: hselasky
2013-01-11 16:10:11 +00:00
daschedule(periph);
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
}
}
/* FALLTHROUGH */
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
case AC_SENT_BDR:
case AC_BUS_RESET:
{
struct ccb_hdr *ccbh;
softc = (struct da_softc *)periph->softc;
/*
* Don't fail on the expected unit attention
* that will occur.
*/
softc->flags |= DA_FLAG_RETRY_UA;
Mechanical change to use <sys/queue.h> macro API instead of fondling implementation details. Created with: sed(1) Reviewed by: md5(1)
2001-02-04 13:13:25 +00:00
LIST_FOREACH(ccbh, &softc->pending_ccbs, periph_links.le)
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
ccbh->ccb_state |= DA_CCB_RETRY_UA;
sys/cam/scsi/scsi_da.c: - Only attempt the closing synchronize cache on a disk if it is still there. - When a device is lost, report the number of outstanding I/Os as they are drained. - When a device is lost, return any unprocessed bios with ENXIO instead of EIO. - Filter asynchronous events, but always allow cam_periph_async() to see them too. Sponsored by: Spectra Logic Corporation
2011-06-14 16:05:00 +00:00
break;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
default:
break;
}
sys/cam/scsi/scsi_da.c: - Only attempt the closing synchronize cache on a disk if it is still there. - When a device is lost, report the number of outstanding I/Os as they are drained. - When a device is lost, return any unprocessed bios with ENXIO instead of EIO. - Filter asynchronous events, but always allow cam_periph_async() to see them too. Sponsored by: Spectra Logic Corporation
2011-06-14 16:05:00 +00:00
cam_periph_async(periph, code, path, arg);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
Move dynamic sysctl(8) variable creation for the cd(4) and da(4) drivers out of cdregister() and daregister(), which are run from interrupt context. The sysctl code does blocking mallocs (M_WAITOK), which causes problems if malloc(9) actually needs to sleep. The eventual fix for this issue will involve moving the CAM probe process inside a kernel thread. For now, though, I have fixed the issue by moving dynamic sysctl variable creation for these two drivers to a task queue running in a kernel thread. The existing task queues (taskqueue_swi and taskqueue_swi_giant) run in software interrupt handlers, which wouldn't fix the problem at hand. So I have created a new task queue, taskqueue_thread, that runs inside a kernel thread. (It also runs outside of Giant -- clients must explicitly acquire and release Giant in their taskqueue functions.) scsi_cd.c: Remove sysctl variable creation code from cdregister(), and move it to a new function, cdsysctlinit(). Queue cdsysctlinit() to the taskqueue_thread taskqueue once we have fully registered the cd(4) driver instance. scsi_da.c: Remove sysctl variable creation code from daregister(), and move it to move it to a new function, dasysctlinit(). Queue dasysctlinit() to the taskqueue_thread taskqueue once we have fully registered the da(4) instance. taskqueue.h: Declare the new taskqueue_thread taskqueue, update some comments. subr_taskqueue.c: Create the new kernel thread taskqueue. This taskqueue runs outside of Giant, so any functions queued to it would need to explicitly acquire/release Giant if they need it. cd.4: Update the cd(4) man page to talk about the minimum command size sysctl/loader tunable. Also note that the changer variables are available as loader tunables as well. da.4: Update the da(4) man page to cover the retry_count, default_timeout and minimum_cmd_size sysctl variables/loader tunables. Remove references to /dev/r???, they aren't used any longer. cd.9: Update the cd(9) man page to describe the CD_Q_10_BYTE_ONLY quirk. taskqueue.9: Update the taskqueue(9) man page to describe the new thread task queue, and the taskqueue_swi_giant queue. MFC after: 3 days
2003-09-03 04:46:28 +00:00
static void
dasysctlinit(void *context, int pending)
{
struct cam_periph *periph;
struct da_softc *softc;
char tmpstr[80], tmpstr2[80];
redfacedly noting a missing part of a patch. Pointy Hat To: Me MFC after: 2 Week X-MFC: 208896
2010-06-07 22:43:37 +00:00
struct ccb_trans_settings cts;
Move dynamic sysctl(8) variable creation for the cd(4) and da(4) drivers out of cdregister() and daregister(), which are run from interrupt context. The sysctl code does blocking mallocs (M_WAITOK), which causes problems if malloc(9) actually needs to sleep. The eventual fix for this issue will involve moving the CAM probe process inside a kernel thread. For now, though, I have fixed the issue by moving dynamic sysctl variable creation for these two drivers to a task queue running in a kernel thread. The existing task queues (taskqueue_swi and taskqueue_swi_giant) run in software interrupt handlers, which wouldn't fix the problem at hand. So I have created a new task queue, taskqueue_thread, that runs inside a kernel thread. (It also runs outside of Giant -- clients must explicitly acquire and release Giant in their taskqueue functions.) scsi_cd.c: Remove sysctl variable creation code from cdregister(), and move it to a new function, cdsysctlinit(). Queue cdsysctlinit() to the taskqueue_thread taskqueue once we have fully registered the cd(4) driver instance. scsi_da.c: Remove sysctl variable creation code from daregister(), and move it to move it to a new function, dasysctlinit(). Queue dasysctlinit() to the taskqueue_thread taskqueue once we have fully registered the da(4) instance. taskqueue.h: Declare the new taskqueue_thread taskqueue, update some comments. subr_taskqueue.c: Create the new kernel thread taskqueue. This taskqueue runs outside of Giant, so any functions queued to it would need to explicitly acquire/release Giant if they need it. cd.4: Update the cd(4) man page to talk about the minimum command size sysctl/loader tunable. Also note that the changer variables are available as loader tunables as well. da.4: Update the da(4) man page to cover the retry_count, default_timeout and minimum_cmd_size sysctl variables/loader tunables. Remove references to /dev/r???, they aren't used any longer. cd.9: Update the cd(9) man page to describe the CD_Q_10_BYTE_ONLY quirk. taskqueue.9: Update the taskqueue(9) man page to describe the new thread task queue, and the taskqueue_swi_giant queue. MFC after: 3 days
2003-09-03 04:46:28 +00:00
periph = (struct cam_periph *)context;
Rearrange how things are done to avoid dereferencing stale pointers in the case of immediate unconfigure after configure. Hold the periph an extra count while we have the task to create sysctl context outstanding so that the periph doesn't go away unexpectedly. Sponsored by: Panasas Reviewed by: scsi@ MFC after: 1 month
2010-06-08 22:46:44 +00:00
/*
* periph was held for us when this task was enqueued
*/
if (periph->flags & CAM_PERIPH_INVALID) {
cam_periph_release(periph);
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
return;
Rearrange how things are done to avoid dereferencing stale pointers in the case of immediate unconfigure after configure. Hold the periph an extra count while we have the task to create sysctl context outstanding so that the periph doesn't go away unexpectedly. Sponsored by: Panasas Reviewed by: scsi@ MFC after: 1 month
2010-06-08 22:46:44 +00:00
}
Move dynamic sysctl(8) variable creation for the cd(4) and da(4) drivers out of cdregister() and daregister(), which are run from interrupt context. The sysctl code does blocking mallocs (M_WAITOK), which causes problems if malloc(9) actually needs to sleep. The eventual fix for this issue will involve moving the CAM probe process inside a kernel thread. For now, though, I have fixed the issue by moving dynamic sysctl variable creation for these two drivers to a task queue running in a kernel thread. The existing task queues (taskqueue_swi and taskqueue_swi_giant) run in software interrupt handlers, which wouldn't fix the problem at hand. So I have created a new task queue, taskqueue_thread, that runs inside a kernel thread. (It also runs outside of Giant -- clients must explicitly acquire and release Giant in their taskqueue functions.) scsi_cd.c: Remove sysctl variable creation code from cdregister(), and move it to a new function, cdsysctlinit(). Queue cdsysctlinit() to the taskqueue_thread taskqueue once we have fully registered the cd(4) driver instance. scsi_da.c: Remove sysctl variable creation code from daregister(), and move it to move it to a new function, dasysctlinit(). Queue dasysctlinit() to the taskqueue_thread taskqueue once we have fully registered the da(4) instance. taskqueue.h: Declare the new taskqueue_thread taskqueue, update some comments. subr_taskqueue.c: Create the new kernel thread taskqueue. This taskqueue runs outside of Giant, so any functions queued to it would need to explicitly acquire/release Giant if they need it. cd.4: Update the cd(4) man page to talk about the minimum command size sysctl/loader tunable. Also note that the changer variables are available as loader tunables as well. da.4: Update the da(4) man page to cover the retry_count, default_timeout and minimum_cmd_size sysctl variables/loader tunables. Remove references to /dev/r???, they aren't used any longer. cd.9: Update the cd(9) man page to describe the CD_Q_10_BYTE_ONLY quirk. taskqueue.9: Update the taskqueue(9) man page to describe the new thread task queue, and the taskqueue_swi_giant queue. MFC after: 3 days
2003-09-03 04:46:28 +00:00
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
softc = (struct da_softc *)periph->softc;
Move dynamic sysctl(8) variable creation for the cd(4) and da(4) drivers out of cdregister() and daregister(), which are run from interrupt context. The sysctl code does blocking mallocs (M_WAITOK), which causes problems if malloc(9) actually needs to sleep. The eventual fix for this issue will involve moving the CAM probe process inside a kernel thread. For now, though, I have fixed the issue by moving dynamic sysctl variable creation for these two drivers to a task queue running in a kernel thread. The existing task queues (taskqueue_swi and taskqueue_swi_giant) run in software interrupt handlers, which wouldn't fix the problem at hand. So I have created a new task queue, taskqueue_thread, that runs inside a kernel thread. (It also runs outside of Giant -- clients must explicitly acquire and release Giant in their taskqueue functions.) scsi_cd.c: Remove sysctl variable creation code from cdregister(), and move it to a new function, cdsysctlinit(). Queue cdsysctlinit() to the taskqueue_thread taskqueue once we have fully registered the cd(4) driver instance. scsi_da.c: Remove sysctl variable creation code from daregister(), and move it to move it to a new function, dasysctlinit(). Queue dasysctlinit() to the taskqueue_thread taskqueue once we have fully registered the da(4) instance. taskqueue.h: Declare the new taskqueue_thread taskqueue, update some comments. subr_taskqueue.c: Create the new kernel thread taskqueue. This taskqueue runs outside of Giant, so any functions queued to it would need to explicitly acquire/release Giant if they need it. cd.4: Update the cd(4) man page to talk about the minimum command size sysctl/loader tunable. Also note that the changer variables are available as loader tunables as well. da.4: Update the da(4) man page to cover the retry_count, default_timeout and minimum_cmd_size sysctl variables/loader tunables. Remove references to /dev/r???, they aren't used any longer. cd.9: Update the cd(9) man page to describe the CD_Q_10_BYTE_ONLY quirk. taskqueue.9: Update the taskqueue(9) man page to describe the new thread task queue, and the taskqueue_swi_giant queue. MFC after: 3 days
2003-09-03 04:46:28 +00:00
snprintf(tmpstr, sizeof(tmpstr), "CAM DA unit %d", periph->unit_number);
snprintf(tmpstr2, sizeof(tmpstr2), "%d", periph->unit_number);
sysctl_ctx_init(&softc->sysctl_ctx);
(dacleanup, dasysctlinit): Defend against calling sysctl_ctx_free on an uninitialized sysctl_ctx, using flag DA_FLAG_SCTX_INIT. This prevents a panic encoutered with some umass units that probe correctly but fail to attach. Same problem, and same fix, as scsi_cd.c rev. 1.86. Reviewed by: njl, ken
2003-10-08 07:12:30 +00:00
softc->flags |= DA_FLAG_SCTX_INIT;
Move dynamic sysctl(8) variable creation for the cd(4) and da(4) drivers out of cdregister() and daregister(), which are run from interrupt context. The sysctl code does blocking mallocs (M_WAITOK), which causes problems if malloc(9) actually needs to sleep. The eventual fix for this issue will involve moving the CAM probe process inside a kernel thread. For now, though, I have fixed the issue by moving dynamic sysctl variable creation for these two drivers to a task queue running in a kernel thread. The existing task queues (taskqueue_swi and taskqueue_swi_giant) run in software interrupt handlers, which wouldn't fix the problem at hand. So I have created a new task queue, taskqueue_thread, that runs inside a kernel thread. (It also runs outside of Giant -- clients must explicitly acquire and release Giant in their taskqueue functions.) scsi_cd.c: Remove sysctl variable creation code from cdregister(), and move it to a new function, cdsysctlinit(). Queue cdsysctlinit() to the taskqueue_thread taskqueue once we have fully registered the cd(4) driver instance. scsi_da.c: Remove sysctl variable creation code from daregister(), and move it to move it to a new function, dasysctlinit(). Queue dasysctlinit() to the taskqueue_thread taskqueue once we have fully registered the da(4) instance. taskqueue.h: Declare the new taskqueue_thread taskqueue, update some comments. subr_taskqueue.c: Create the new kernel thread taskqueue. This taskqueue runs outside of Giant, so any functions queued to it would need to explicitly acquire/release Giant if they need it. cd.4: Update the cd(4) man page to talk about the minimum command size sysctl/loader tunable. Also note that the changer variables are available as loader tunables as well. da.4: Update the da(4) man page to cover the retry_count, default_timeout and minimum_cmd_size sysctl variables/loader tunables. Remove references to /dev/r???, they aren't used any longer. cd.9: Update the cd(9) man page to describe the CD_Q_10_BYTE_ONLY quirk. taskqueue.9: Update the taskqueue(9) man page to describe the new thread task queue, and the taskqueue_swi_giant queue. MFC after: 3 days
2003-09-03 04:46:28 +00:00
softc->sysctl_tree = SYSCTL_ADD_NODE(&softc->sysctl_ctx,
SYSCTL_STATIC_CHILDREN(_kern_cam_da), OID_AUTO, tmpstr2,
CTLFLAG_RD, 0, tmpstr);
if (softc->sysctl_tree == NULL) {
printf("dasysctlinit: unable to allocate sysctl tree\n");
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
cam_periph_release(periph);
Move dynamic sysctl(8) variable creation for the cd(4) and da(4) drivers out of cdregister() and daregister(), which are run from interrupt context. The sysctl code does blocking mallocs (M_WAITOK), which causes problems if malloc(9) actually needs to sleep. The eventual fix for this issue will involve moving the CAM probe process inside a kernel thread. For now, though, I have fixed the issue by moving dynamic sysctl variable creation for these two drivers to a task queue running in a kernel thread. The existing task queues (taskqueue_swi and taskqueue_swi_giant) run in software interrupt handlers, which wouldn't fix the problem at hand. So I have created a new task queue, taskqueue_thread, that runs inside a kernel thread. (It also runs outside of Giant -- clients must explicitly acquire and release Giant in their taskqueue functions.) scsi_cd.c: Remove sysctl variable creation code from cdregister(), and move it to a new function, cdsysctlinit(). Queue cdsysctlinit() to the taskqueue_thread taskqueue once we have fully registered the cd(4) driver instance. scsi_da.c: Remove sysctl variable creation code from daregister(), and move it to move it to a new function, dasysctlinit(). Queue dasysctlinit() to the taskqueue_thread taskqueue once we have fully registered the da(4) instance. taskqueue.h: Declare the new taskqueue_thread taskqueue, update some comments. subr_taskqueue.c: Create the new kernel thread taskqueue. This taskqueue runs outside of Giant, so any functions queued to it would need to explicitly acquire/release Giant if they need it. cd.4: Update the cd(4) man page to talk about the minimum command size sysctl/loader tunable. Also note that the changer variables are available as loader tunables as well. da.4: Update the da(4) man page to cover the retry_count, default_timeout and minimum_cmd_size sysctl variables/loader tunables. Remove references to /dev/r???, they aren't used any longer. cd.9: Update the cd(9) man page to describe the CD_Q_10_BYTE_ONLY quirk. taskqueue.9: Update the taskqueue(9) man page to describe the new thread task queue, and the taskqueue_swi_giant queue. MFC after: 3 days
2003-09-03 04:46:28 +00:00
return;
}
/*
Do a minor amount of stylifying. Also, get a Fibre Channel WWPN if one exists for a da unit and create a sysctl OID for it.
2010-06-07 17:41:34 +00:00
* Now register the sysctl handler, so the user can change the value on
Move dynamic sysctl(8) variable creation for the cd(4) and da(4) drivers out of cdregister() and daregister(), which are run from interrupt context. The sysctl code does blocking mallocs (M_WAITOK), which causes problems if malloc(9) actually needs to sleep. The eventual fix for this issue will involve moving the CAM probe process inside a kernel thread. For now, though, I have fixed the issue by moving dynamic sysctl variable creation for these two drivers to a task queue running in a kernel thread. The existing task queues (taskqueue_swi and taskqueue_swi_giant) run in software interrupt handlers, which wouldn't fix the problem at hand. So I have created a new task queue, taskqueue_thread, that runs inside a kernel thread. (It also runs outside of Giant -- clients must explicitly acquire and release Giant in their taskqueue functions.) scsi_cd.c: Remove sysctl variable creation code from cdregister(), and move it to a new function, cdsysctlinit(). Queue cdsysctlinit() to the taskqueue_thread taskqueue once we have fully registered the cd(4) driver instance. scsi_da.c: Remove sysctl variable creation code from daregister(), and move it to move it to a new function, dasysctlinit(). Queue dasysctlinit() to the taskqueue_thread taskqueue once we have fully registered the da(4) instance. taskqueue.h: Declare the new taskqueue_thread taskqueue, update some comments. subr_taskqueue.c: Create the new kernel thread taskqueue. This taskqueue runs outside of Giant, so any functions queued to it would need to explicitly acquire/release Giant if they need it. cd.4: Update the cd(4) man page to talk about the minimum command size sysctl/loader tunable. Also note that the changer variables are available as loader tunables as well. da.4: Update the da(4) man page to cover the retry_count, default_timeout and minimum_cmd_size sysctl variables/loader tunables. Remove references to /dev/r???, they aren't used any longer. cd.9: Update the cd(9) man page to describe the CD_Q_10_BYTE_ONLY quirk. taskqueue.9: Update the taskqueue(9) man page to describe the new thread task queue, and the taskqueue_swi_giant queue. MFC after: 3 days
2003-09-03 04:46:28 +00:00
* the fly.
*/
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
SYSCTL_ADD_PROC(&softc->sysctl_ctx, SYSCTL_CHILDREN(softc->sysctl_tree),
Make delete method set via kern.cam.da.X.delete_method persistent. This allows to set delete method via tunable, before device capabilities are known. Also allow ZERO method for devices not reporting LBP, if user explicitly requests it -- it may be useful if storage supports compression and WRITE SAME, but does not support UNMAP. MFC after: 2 weeks
2015-10-11 18:26:06 +00:00
OID_AUTO, "delete_method", CTLTYPE_STRING | CTLFLAG_RWTUN,
Updates delete_method sysctl changes to always maintain disk d_flags DISKFLAG_CANDELETE. While this change makes this layer consistent other layers such as UFS and ZFS BIO_DELETE support may not notice any change made manually via these device sysctls until the device is reopened via a mount. Also corrected var order in dadeletemethodsysctl PR: kern/169801 Reviewed by: pjd (mentor) Approved by: mav MFC after: 2 weeks
2013-01-10 11:57:46 +00:00
softc, 0, dadeletemethodsysctl, "A",
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
"BIO_DELETE execution method");
Teach GEOM and CAM about the difference between the max "size" of r/w and delete requests. sys/geom/geom_disk.h: - Added d_delmaxsize which represents the maximum size of individual device delete requests in bytes. This can be used by devices to inform geom of their size limitations regarding delete operations which are generally different from the read / write limits as data is not usually transferred from the host to physical device. sys/geom/geom_disk.c: - Use new d_delmaxsize to calculate the size of chunks passed through to the underlying strategy during deletes instead of using read / write optimised values. This defaults to d_maxsize if unset (0). - Moved d_maxsize default up so it can be used to default d_delmaxsize sys/cam/ata/ata_da.c: - Added d_delmaxsize calculations for TRIM and CFA sys/cam/scsi/scsi_da.c: - Added re-calculation of d_delmaxsize whenever delete_method is set. - Added kern.cam.da.X.delete_max sysctl which allows the max size for delete requests to be limited. This is useful in preventing timeouts on devices who's delete methods are slow. It should be noted that this limit is reset then the device delete method is changed and that it can only be lowered not increased from the device max. Reviewed by: mav Approved by: pjd (mentor)
2013-04-26 16:22:54 +00:00
SYSCTL_ADD_PROC(&softc->sysctl_ctx, SYSCTL_CHILDREN(softc->sysctl_tree),
OID_AUTO, "delete_max", CTLTYPE_U64 | CTLFLAG_RW,
softc, 0, dadeletemaxsysctl, "Q",
"Maximum BIO_DELETE size");
Do a minor amount of stylifying. Also, get a Fibre Channel WWPN if one exists for a da unit and create a sysctl OID for it.
2010-06-07 17:41:34 +00:00
SYSCTL_ADD_PROC(&softc->sysctl_ctx, SYSCTL_CHILDREN(softc->sysctl_tree),
Move dynamic sysctl(8) variable creation for the cd(4) and da(4) drivers out of cdregister() and daregister(), which are run from interrupt context. The sysctl code does blocking mallocs (M_WAITOK), which causes problems if malloc(9) actually needs to sleep. The eventual fix for this issue will involve moving the CAM probe process inside a kernel thread. For now, though, I have fixed the issue by moving dynamic sysctl variable creation for these two drivers to a task queue running in a kernel thread. The existing task queues (taskqueue_swi and taskqueue_swi_giant) run in software interrupt handlers, which wouldn't fix the problem at hand. So I have created a new task queue, taskqueue_thread, that runs inside a kernel thread. (It also runs outside of Giant -- clients must explicitly acquire and release Giant in their taskqueue functions.) scsi_cd.c: Remove sysctl variable creation code from cdregister(), and move it to a new function, cdsysctlinit(). Queue cdsysctlinit() to the taskqueue_thread taskqueue once we have fully registered the cd(4) driver instance. scsi_da.c: Remove sysctl variable creation code from daregister(), and move it to move it to a new function, dasysctlinit(). Queue dasysctlinit() to the taskqueue_thread taskqueue once we have fully registered the da(4) instance. taskqueue.h: Declare the new taskqueue_thread taskqueue, update some comments. subr_taskqueue.c: Create the new kernel thread taskqueue. This taskqueue runs outside of Giant, so any functions queued to it would need to explicitly acquire/release Giant if they need it. cd.4: Update the cd(4) man page to talk about the minimum command size sysctl/loader tunable. Also note that the changer variables are available as loader tunables as well. da.4: Update the da(4) man page to cover the retry_count, default_timeout and minimum_cmd_size sysctl variables/loader tunables. Remove references to /dev/r???, they aren't used any longer. cd.9: Update the cd(9) man page to describe the CD_Q_10_BYTE_ONLY quirk. taskqueue.9: Update the taskqueue(9) man page to describe the new thread task queue, and the taskqueue_swi_giant queue. MFC after: 3 days
2003-09-03 04:46:28 +00:00
OID_AUTO, "minimum_cmd_size", CTLTYPE_INT | CTLFLAG_RW,
&softc->minimum_cmd_size, 0, dacmdsizesysctl, "I",
"Minimum CDB size");
Adds the ability to enable / disable sorting of BIO requests queued within CAM. This can significantly improve performance particularly for SSDs which don't suffer from seek latencies. The sysctl / tunable kern.cam.sort_io_queues provides the systems default setting where:- 0 = queued BIOs are NOT sorted 1 = queued BIOs are sorted (default) Each device gets its own sysctl kern.cam.<type>.<id>.sort_io_queue Valid values are:- -1 = use system default (default) 0 = queued BIOs are NOT sorted 1 = queued BIOs are sorted Note: Additional patch will look to add automatic use of none sorted queues for none rotating media e.g. SSD's Reviewed by: scottl Approved by: pjd (mentor) MFC after: 2 weeks
2013-03-29 22:58:15 +00:00
SYSCTL_ADD_INT(&softc->sysctl_ctx, SYSCTL_CHILDREN(softc->sysctl_tree),
OID_AUTO, "sort_io_queue", CTLFLAG_RW, &softc->sort_io_queue, 0,
"Sort IO queue to try and optimise disk access patterns");
Move dynamic sysctl(8) variable creation for the cd(4) and da(4) drivers out of cdregister() and daregister(), which are run from interrupt context. The sysctl code does blocking mallocs (M_WAITOK), which causes problems if malloc(9) actually needs to sleep. The eventual fix for this issue will involve moving the CAM probe process inside a kernel thread. For now, though, I have fixed the issue by moving dynamic sysctl variable creation for these two drivers to a task queue running in a kernel thread. The existing task queues (taskqueue_swi and taskqueue_swi_giant) run in software interrupt handlers, which wouldn't fix the problem at hand. So I have created a new task queue, taskqueue_thread, that runs inside a kernel thread. (It also runs outside of Giant -- clients must explicitly acquire and release Giant in their taskqueue functions.) scsi_cd.c: Remove sysctl variable creation code from cdregister(), and move it to a new function, cdsysctlinit(). Queue cdsysctlinit() to the taskqueue_thread taskqueue once we have fully registered the cd(4) driver instance. scsi_da.c: Remove sysctl variable creation code from daregister(), and move it to move it to a new function, dasysctlinit(). Queue dasysctlinit() to the taskqueue_thread taskqueue once we have fully registered the da(4) instance. taskqueue.h: Declare the new taskqueue_thread taskqueue, update some comments. subr_taskqueue.c: Create the new kernel thread taskqueue. This taskqueue runs outside of Giant, so any functions queued to it would need to explicitly acquire/release Giant if they need it. cd.4: Update the cd(4) man page to talk about the minimum command size sysctl/loader tunable. Also note that the changer variables are available as loader tunables as well. da.4: Update the da(4) man page to cover the retry_count, default_timeout and minimum_cmd_size sysctl variables/loader tunables. Remove references to /dev/r???, they aren't used any longer. cd.9: Update the cd(9) man page to describe the CD_Q_10_BYTE_ONLY quirk. taskqueue.9: Update the taskqueue(9) man page to describe the new thread task queue, and the taskqueue_swi_giant queue. MFC after: 3 days
2003-09-03 04:46:28 +00:00
Fix a race condition in CAM peripheral free handling, locking in the CAM XPT bus traversal code, and a number of other periph level issues. cam_periph.h, cam_periph.c: Modify cam_periph_acquire() to test the CAM_PERIPH_INVALID flag prior to allowing a reference count to be gained on a peripheral. Callers of this function will receive CAM_REQ_CMP_ERR status in the situation of attempting to reference an invalidated periph. This guarantees that a peripheral scheduled for a deferred free will not be accessed during its wait for destruction. Panic during attempts to drop a reference count on a peripheral that already has a zero reference count. In cam_periph_list(), use a local sbuf with SBUF_FIXEDLEN set so that mallocs do not occur while the xpt topology lock is held, regardless of the allocation policy of the passed in sbuf. Add a new routine, cam_periph_release_locked_buses(), that can be called when the caller already holds the CAM topology lock. Add some extra debugging for duplicate peripheral allocations in cam_periph_alloc(). Treat CAM_DEV_NOT_THERE much the same as a selection timeout (AC_LOST_DEVICE is emitted), but forgo retries. cam_xpt.c: Revamp the way the EDT traversal code does locking and reference counting. This was broken, since it assumed that the EDT would not change during traversal, but that assumption is no longer valid. So, to prevent devices from going away while we traverse the EDT, make sure we properly lock everything and hold references on devices that we are using. The two peripheral driver traversal routines should be examined. xptpdperiphtraverse() holds the topology lock for the entire time it runs. xptperiphtraverse() is now locked properly, but only holds the topology lock while it is traversing the list, and not while the traversal function is running. The bus locking code in xptbustraverse() should also be revisited at a later time, since it is complex and should probably be simplified. scsi_da.c: Pay attention to the return value from cam_periph_acquire(). Return 0 always from daclose() even if the disk is now gone. Add some rudimentary error injection support. scsi_sg.c: Fix reference counting in the sg(4) driver. The sg driver was calling cam_periph_release() on close, but never called cam_periph_acquire() (which increments the reference count) on open. The periph code correctly complained that the sg(4) driver was trying to decrement the refcount when it was already 0. Sponsored by: Spectra Logic MFC after: 2 weeks
2012-01-12 00:41:48 +00:00
SYSCTL_ADD_INT(&softc->sysctl_ctx,
SYSCTL_CHILDREN(softc->sysctl_tree),
OID_AUTO,
"error_inject",
CTLFLAG_RW,
&softc->error_inject,
0,
"error_inject leaf");
Do a minor amount of stylifying. Also, get a Fibre Channel WWPN if one exists for a da unit and create a sysctl OID for it.
2010-06-07 17:41:34 +00:00
/*
* Add some addressing info.
*/
memset(&cts, 0, sizeof (cts));
Remove several uses of numeric priorities from immediate CCB setups.
2012-10-27 09:40:29 +00:00
xpt_setup_ccb(&cts.ccb_h, periph->path, CAM_PRIORITY_NONE);
Do a minor amount of stylifying. Also, get a Fibre Channel WWPN if one exists for a da unit and create a sysctl OID for it.
2010-06-07 17:41:34 +00:00
cts.ccb_h.func_code = XPT_GET_TRAN_SETTINGS;
cts.type = CTS_TYPE_CURRENT_SETTINGS;
cam_periph_lock(periph);
xpt_action((union ccb *)&cts);
cam_periph_unlock(periph);
if (cts.ccb_h.status != CAM_REQ_CMP) {
cam_periph_release(periph);
return;
}
if (cts.protocol == PROTO_SCSI && cts.transport == XPORT_FC) {
struct ccb_trans_settings_fc *fc = &cts.xport_specific.fc;
if (fc->valid & CTS_FC_VALID_WWPN) {
softc->wwpn = fc->wwpn;
sysctl(8) should use the CTLTYPE to determine the type of data when reading. (This was already done for writing to a sysctl). This requires all SYSCTL setups to specify a type. Most of them are now checked at compile-time. Remove SYSCTL_*X* sysctl additions as the print being in hex should be controlled by the -x flag to sysctl(8). Succested by: bde
2011-01-19 17:04:07 +00:00
SYSCTL_ADD_UQUAD(&softc->sysctl_ctx,
Do a minor amount of stylifying. Also, get a Fibre Channel WWPN if one exists for a da unit and create a sysctl OID for it.
2010-06-07 17:41:34 +00:00
SYSCTL_CHILDREN(softc->sysctl_tree),
Add a 64-bit hex-printed sysctl(9) since there is at least one place in the code that wanted it. It is named X64 rather than XQUAD since the quad name is a historical abomination that should not be perpetuated.
2011-01-13 18:20:33 +00:00
OID_AUTO, "wwpn", CTLFLAG_RD,
Do a minor amount of stylifying. Also, get a Fibre Channel WWPN if one exists for a da unit and create a sysctl OID for it.
2010-06-07 17:41:34 +00:00
&softc->wwpn, "World Wide Port Name");
}
}
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
cam_periph_release(periph);
Move dynamic sysctl(8) variable creation for the cd(4) and da(4) drivers out of cdregister() and daregister(), which are run from interrupt context. The sysctl code does blocking mallocs (M_WAITOK), which causes problems if malloc(9) actually needs to sleep. The eventual fix for this issue will involve moving the CAM probe process inside a kernel thread. For now, though, I have fixed the issue by moving dynamic sysctl variable creation for these two drivers to a task queue running in a kernel thread. The existing task queues (taskqueue_swi and taskqueue_swi_giant) run in software interrupt handlers, which wouldn't fix the problem at hand. So I have created a new task queue, taskqueue_thread, that runs inside a kernel thread. (It also runs outside of Giant -- clients must explicitly acquire and release Giant in their taskqueue functions.) scsi_cd.c: Remove sysctl variable creation code from cdregister(), and move it to a new function, cdsysctlinit(). Queue cdsysctlinit() to the taskqueue_thread taskqueue once we have fully registered the cd(4) driver instance. scsi_da.c: Remove sysctl variable creation code from daregister(), and move it to move it to a new function, dasysctlinit(). Queue dasysctlinit() to the taskqueue_thread taskqueue once we have fully registered the da(4) instance. taskqueue.h: Declare the new taskqueue_thread taskqueue, update some comments. subr_taskqueue.c: Create the new kernel thread taskqueue. This taskqueue runs outside of Giant, so any functions queued to it would need to explicitly acquire/release Giant if they need it. cd.4: Update the cd(4) man page to talk about the minimum command size sysctl/loader tunable. Also note that the changer variables are available as loader tunables as well. da.4: Update the da(4) man page to cover the retry_count, default_timeout and minimum_cmd_size sysctl variables/loader tunables. Remove references to /dev/r???, they aren't used any longer. cd.9: Update the cd(9) man page to describe the CD_Q_10_BYTE_ONLY quirk. taskqueue.9: Update the taskqueue(9) man page to describe the new thread task queue, and the taskqueue_swi_giant queue. MFC after: 3 days
2003-09-03 04:46:28 +00:00
}
Teach GEOM and CAM about the difference between the max "size" of r/w and delete requests. sys/geom/geom_disk.h: - Added d_delmaxsize which represents the maximum size of individual device delete requests in bytes. This can be used by devices to inform geom of their size limitations regarding delete operations which are generally different from the read / write limits as data is not usually transferred from the host to physical device. sys/geom/geom_disk.c: - Use new d_delmaxsize to calculate the size of chunks passed through to the underlying strategy during deletes instead of using read / write optimised values. This defaults to d_maxsize if unset (0). - Moved d_maxsize default up so it can be used to default d_delmaxsize sys/cam/ata/ata_da.c: - Added d_delmaxsize calculations for TRIM and CFA sys/cam/scsi/scsi_da.c: - Added re-calculation of d_delmaxsize whenever delete_method is set. - Added kern.cam.da.X.delete_max sysctl which allows the max size for delete requests to be limited. This is useful in preventing timeouts on devices who's delete methods are slow. It should be noted that this limit is reset then the device delete method is changed and that it can only be lowered not increased from the device max. Reviewed by: mav Approved by: pjd (mentor)
2013-04-26 16:22:54 +00:00
static int
dadeletemaxsysctl(SYSCTL_HANDLER_ARGS)
{
int error;
uint64_t value;
struct da_softc *softc;
softc = (struct da_softc *)arg1;
value = softc->disk->d_delmaxsize;
error = sysctl_handle_64(oidp, &value, 0, req);
if ((error != 0) || (req->newptr == NULL))
return (error);
/* only accept values smaller than the calculated value */
Bump disk(9) ABI version to signify the addition of d_delmaxsize by r249940. Ensure that d_delmaxsize is always set, removing init to 0 which could cause future issues if use cases change. Allow kern.cam.da.X.delete_max (which maps to d_delmaxsize) to be increased up to the calculated max after being reduced. MFC after: 1 day X-MFC-With: r249940
2013-07-03 23:46:30 +00:00
if (value > dadeletemaxsize(softc, softc->delete_method)) {
Teach GEOM and CAM about the difference between the max "size" of r/w and delete requests. sys/geom/geom_disk.h: - Added d_delmaxsize which represents the maximum size of individual device delete requests in bytes. This can be used by devices to inform geom of their size limitations regarding delete operations which are generally different from the read / write limits as data is not usually transferred from the host to physical device. sys/geom/geom_disk.c: - Use new d_delmaxsize to calculate the size of chunks passed through to the underlying strategy during deletes instead of using read / write optimised values. This defaults to d_maxsize if unset (0). - Moved d_maxsize default up so it can be used to default d_delmaxsize sys/cam/ata/ata_da.c: - Added d_delmaxsize calculations for TRIM and CFA sys/cam/scsi/scsi_da.c: - Added re-calculation of d_delmaxsize whenever delete_method is set. - Added kern.cam.da.X.delete_max sysctl which allows the max size for delete requests to be limited. This is useful in preventing timeouts on devices who's delete methods are slow. It should be noted that this limit is reset then the device delete method is changed and that it can only be lowered not increased from the device max. Reviewed by: mav Approved by: pjd (mentor)
2013-04-26 16:22:54 +00:00
return (EINVAL);
}
softc->disk->d_delmaxsize = value;
return (0);
}
Fix ATAPI/USB/Firewire CDROM drive handling in cd(4) and hopefully fix a number of related problems along the way. - Automatically detect CDROM drives that can't handle 6 byte mode sense and mode select, and adjust our command size accordingly. We have to handle this in the cd(4) driver (where the buffers are allocated), since the parameter list length is different for the 6 and 10 byte mode sense commands. - Remove MODE_SENSE and MODE_SELECT translation removed in ATAPICAM and in the umass(4) driver, since there's no way for that to work properly. - Add a quirk entry for CDROM drives that just hang when they get a 6 byte mode sense or mode select. The reason for the quirk must be documented in a PR, and all quirks must be approved by ken@FreeBSD.org. This is to make sure that we fully understand why each quirk is needed. Once the CAM_NEW_TRAN_CODE is finished, we should be able to remove any such quirks, since we'll know what protocol the drive speaks (SCSI, ATAPI, etc.) and therefore whether we should use 6 or 10 byte mode sense/select commands. - Change the way the da(4) handles the no_6_byte sysctl. There is now a per-drive sysctl to set the minimum command size for that particular disk. (Since you could have multiple disks with multiple requirements in one system.) - Loader tunable support for all the sysctls in the da(4) and cd(4) drivers. - Add a CDIOCCLOSE ioctl for cd(4) (bde pointed this out a long time ago). - Add a media validation routine (cdcheckmedia()) to the cd(4) driver, to fix some problems bde pointed out a long time ago. We now allow open() to succeed no matter what, but if we don't detect valid media, the user can only issue CDIOCCLOSE or CDIOCEJECT ioctls. - The media validation routine also reads the table of contents off the drive. We use the table of contents to implement the CDIOCPLAYTRACKS ioctl using the PLAY AUDIO MSF command. The PLAY AUDIO TRACK INDEX command that we previously used was deprecated after SCSI-2. It works in every SCSI CDROM I've tried, but doesn't seem to work on ATAPI CDROM drives. We still use the play audio track index command if we don't have a valid TOC, but I suppose it'll fail anyway in that case. - Add _len() versions of scsi_mode_sense() and scsi_mode_select() so that we can specify the minimum command length. - Fix a couple of formatting problems in the sense printing code. MFC after: 4 weeks
2003-02-21 06:19:38 +00:00
static int
dacmdsizesysctl(SYSCTL_HANDLER_ARGS)
{
int error, value;
value = *(int *)arg1;
error = sysctl_handle_int(oidp, &value, 0, req);
if ((error != 0)
|| (req->newptr == NULL))
return (error);
/*
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
* Acceptable values here are 6, 10, 12 or 16.
Fix ATAPI/USB/Firewire CDROM drive handling in cd(4) and hopefully fix a number of related problems along the way. - Automatically detect CDROM drives that can't handle 6 byte mode sense and mode select, and adjust our command size accordingly. We have to handle this in the cd(4) driver (where the buffers are allocated), since the parameter list length is different for the 6 and 10 byte mode sense commands. - Remove MODE_SENSE and MODE_SELECT translation removed in ATAPICAM and in the umass(4) driver, since there's no way for that to work properly. - Add a quirk entry for CDROM drives that just hang when they get a 6 byte mode sense or mode select. The reason for the quirk must be documented in a PR, and all quirks must be approved by ken@FreeBSD.org. This is to make sure that we fully understand why each quirk is needed. Once the CAM_NEW_TRAN_CODE is finished, we should be able to remove any such quirks, since we'll know what protocol the drive speaks (SCSI, ATAPI, etc.) and therefore whether we should use 6 or 10 byte mode sense/select commands. - Change the way the da(4) handles the no_6_byte sysctl. There is now a per-drive sysctl to set the minimum command size for that particular disk. (Since you could have multiple disks with multiple requirements in one system.) - Loader tunable support for all the sysctls in the da(4) and cd(4) drivers. - Add a CDIOCCLOSE ioctl for cd(4) (bde pointed this out a long time ago). - Add a media validation routine (cdcheckmedia()) to the cd(4) driver, to fix some problems bde pointed out a long time ago. We now allow open() to succeed no matter what, but if we don't detect valid media, the user can only issue CDIOCCLOSE or CDIOCEJECT ioctls. - The media validation routine also reads the table of contents off the drive. We use the table of contents to implement the CDIOCPLAYTRACKS ioctl using the PLAY AUDIO MSF command. The PLAY AUDIO TRACK INDEX command that we previously used was deprecated after SCSI-2. It works in every SCSI CDROM I've tried, but doesn't seem to work on ATAPI CDROM drives. We still use the play audio track index command if we don't have a valid TOC, but I suppose it'll fail anyway in that case. - Add _len() versions of scsi_mode_sense() and scsi_mode_select() so that we can specify the minimum command length. - Fix a couple of formatting problems in the sense printing code. MFC after: 4 weeks
2003-02-21 06:19:38 +00:00
*/
if (value < 6)
value = 6;
else if ((value > 6)
&& (value <= 10))
value = 10;
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
else if ((value > 10)
&& (value <= 12))
Fix ATAPI/USB/Firewire CDROM drive handling in cd(4) and hopefully fix a number of related problems along the way. - Automatically detect CDROM drives that can't handle 6 byte mode sense and mode select, and adjust our command size accordingly. We have to handle this in the cd(4) driver (where the buffers are allocated), since the parameter list length is different for the 6 and 10 byte mode sense commands. - Remove MODE_SENSE and MODE_SELECT translation removed in ATAPICAM and in the umass(4) driver, since there's no way for that to work properly. - Add a quirk entry for CDROM drives that just hang when they get a 6 byte mode sense or mode select. The reason for the quirk must be documented in a PR, and all quirks must be approved by ken@FreeBSD.org. This is to make sure that we fully understand why each quirk is needed. Once the CAM_NEW_TRAN_CODE is finished, we should be able to remove any such quirks, since we'll know what protocol the drive speaks (SCSI, ATAPI, etc.) and therefore whether we should use 6 or 10 byte mode sense/select commands. - Change the way the da(4) handles the no_6_byte sysctl. There is now a per-drive sysctl to set the minimum command size for that particular disk. (Since you could have multiple disks with multiple requirements in one system.) - Loader tunable support for all the sysctls in the da(4) and cd(4) drivers. - Add a CDIOCCLOSE ioctl for cd(4) (bde pointed this out a long time ago). - Add a media validation routine (cdcheckmedia()) to the cd(4) driver, to fix some problems bde pointed out a long time ago. We now allow open() to succeed no matter what, but if we don't detect valid media, the user can only issue CDIOCCLOSE or CDIOCEJECT ioctls. - The media validation routine also reads the table of contents off the drive. We use the table of contents to implement the CDIOCPLAYTRACKS ioctl using the PLAY AUDIO MSF command. The PLAY AUDIO TRACK INDEX command that we previously used was deprecated after SCSI-2. It works in every SCSI CDROM I've tried, but doesn't seem to work on ATAPI CDROM drives. We still use the play audio track index command if we don't have a valid TOC, but I suppose it'll fail anyway in that case. - Add _len() versions of scsi_mode_sense() and scsi_mode_select() so that we can specify the minimum command length. - Fix a couple of formatting problems in the sense printing code. MFC after: 4 weeks
2003-02-21 06:19:38 +00:00
value = 12;
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
else if (value > 12)
value = 16;
Fix ATAPI/USB/Firewire CDROM drive handling in cd(4) and hopefully fix a number of related problems along the way. - Automatically detect CDROM drives that can't handle 6 byte mode sense and mode select, and adjust our command size accordingly. We have to handle this in the cd(4) driver (where the buffers are allocated), since the parameter list length is different for the 6 and 10 byte mode sense commands. - Remove MODE_SENSE and MODE_SELECT translation removed in ATAPICAM and in the umass(4) driver, since there's no way for that to work properly. - Add a quirk entry for CDROM drives that just hang when they get a 6 byte mode sense or mode select. The reason for the quirk must be documented in a PR, and all quirks must be approved by ken@FreeBSD.org. This is to make sure that we fully understand why each quirk is needed. Once the CAM_NEW_TRAN_CODE is finished, we should be able to remove any such quirks, since we'll know what protocol the drive speaks (SCSI, ATAPI, etc.) and therefore whether we should use 6 or 10 byte mode sense/select commands. - Change the way the da(4) handles the no_6_byte sysctl. There is now a per-drive sysctl to set the minimum command size for that particular disk. (Since you could have multiple disks with multiple requirements in one system.) - Loader tunable support for all the sysctls in the da(4) and cd(4) drivers. - Add a CDIOCCLOSE ioctl for cd(4) (bde pointed this out a long time ago). - Add a media validation routine (cdcheckmedia()) to the cd(4) driver, to fix some problems bde pointed out a long time ago. We now allow open() to succeed no matter what, but if we don't detect valid media, the user can only issue CDIOCCLOSE or CDIOCEJECT ioctls. - The media validation routine also reads the table of contents off the drive. We use the table of contents to implement the CDIOCPLAYTRACKS ioctl using the PLAY AUDIO MSF command. The PLAY AUDIO TRACK INDEX command that we previously used was deprecated after SCSI-2. It works in every SCSI CDROM I've tried, but doesn't seem to work on ATAPI CDROM drives. We still use the play audio track index command if we don't have a valid TOC, but I suppose it'll fail anyway in that case. - Add _len() versions of scsi_mode_sense() and scsi_mode_select() so that we can specify the minimum command length. - Fix a couple of formatting problems in the sense printing code. MFC after: 4 weeks
2003-02-21 06:19:38 +00:00
*(int *)arg1 = value;
return (0);
}
Removed unneeded tests in dadeletemethodset changing it to return void Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 15:31:52 +00:00
static void
Updates delete_method sysctl changes to always maintain disk d_flags DISKFLAG_CANDELETE. While this change makes this layer consistent other layers such as UFS and ZFS BIO_DELETE support may not notice any change made manually via these device sysctls until the device is reopened via a mount. Also corrected var order in dadeletemethodsysctl PR: kern/169801 Reviewed by: pjd (mentor) Approved by: mav MFC after: 2 weeks
2013-01-10 11:57:46 +00:00
dadeletemethodset(struct da_softc *softc, da_delete_methods delete_method)
{
softc->delete_method = delete_method;
Teach GEOM and CAM about the difference between the max "size" of r/w and delete requests. sys/geom/geom_disk.h: - Added d_delmaxsize which represents the maximum size of individual device delete requests in bytes. This can be used by devices to inform geom of their size limitations regarding delete operations which are generally different from the read / write limits as data is not usually transferred from the host to physical device. sys/geom/geom_disk.c: - Use new d_delmaxsize to calculate the size of chunks passed through to the underlying strategy during deletes instead of using read / write optimised values. This defaults to d_maxsize if unset (0). - Moved d_maxsize default up so it can be used to default d_delmaxsize sys/cam/ata/ata_da.c: - Added d_delmaxsize calculations for TRIM and CFA sys/cam/scsi/scsi_da.c: - Added re-calculation of d_delmaxsize whenever delete_method is set. - Added kern.cam.da.X.delete_max sysctl which allows the max size for delete requests to be limited. This is useful in preventing timeouts on devices who's delete methods are slow. It should be noted that this limit is reset then the device delete method is changed and that it can only be lowered not increased from the device max. Reviewed by: mav Approved by: pjd (mentor)
2013-04-26 16:22:54 +00:00
softc->disk->d_delmaxsize = dadeletemaxsize(softc, delete_method);
Const-ify the new da_delete_functions. Remove a redundant sanity check Submitted by: Steven Hartland Obtained from: Netflix MFC after: 3 days
2013-07-12 23:20:11 +00:00
softc->delete_func = da_delete_functions[delete_method];
Updates delete_method sysctl changes to always maintain disk d_flags DISKFLAG_CANDELETE. While this change makes this layer consistent other layers such as UFS and ZFS BIO_DELETE support may not notice any change made manually via these device sysctls until the device is reopened via a mount. Also corrected var order in dadeletemethodsysctl PR: kern/169801 Reviewed by: pjd (mentor) Approved by: mav MFC after: 2 weeks
2013-01-10 11:57:46 +00:00
if (softc->delete_method > DA_DELETE_DISABLE)
softc->disk->d_flags |= DISKFLAG_CANDELETE;
else
softc->disk->d_flags &= ~DISKFLAG_CANDELETE;
}
Teach GEOM and CAM about the difference between the max "size" of r/w and delete requests. sys/geom/geom_disk.h: - Added d_delmaxsize which represents the maximum size of individual device delete requests in bytes. This can be used by devices to inform geom of their size limitations regarding delete operations which are generally different from the read / write limits as data is not usually transferred from the host to physical device. sys/geom/geom_disk.c: - Use new d_delmaxsize to calculate the size of chunks passed through to the underlying strategy during deletes instead of using read / write optimised values. This defaults to d_maxsize if unset (0). - Moved d_maxsize default up so it can be used to default d_delmaxsize sys/cam/ata/ata_da.c: - Added d_delmaxsize calculations for TRIM and CFA sys/cam/scsi/scsi_da.c: - Added re-calculation of d_delmaxsize whenever delete_method is set. - Added kern.cam.da.X.delete_max sysctl which allows the max size for delete requests to be limited. This is useful in preventing timeouts on devices who's delete methods are slow. It should be noted that this limit is reset then the device delete method is changed and that it can only be lowered not increased from the device max. Reviewed by: mav Approved by: pjd (mentor)
2013-04-26 16:22:54 +00:00
static off_t
dadeletemaxsize(struct da_softc *softc, da_delete_methods delete_method)
{
off_t sectors;
switch(delete_method) {
case DA_DELETE_UNMAP:
Mostly revert r260267 and hopefully really fix the original problem. The latest draft of SBC-3 tells: "A MAXIMUM UNMAP LBA COUNT field set to a non-zero value indicates the maximum number of LBAs that may be unmapped by an UNMAP command." To me it does not sound like that limit is set per single descriptor, but rather per all command. And I have at least one device that behaves exactly that way. This patch fixes the problem there. MFC after: 1 week
2014-01-22 22:19:53 +00:00
sectors = (off_t)softc->unmap_max_lba;
Teach GEOM and CAM about the difference between the max "size" of r/w and delete requests. sys/geom/geom_disk.h: - Added d_delmaxsize which represents the maximum size of individual device delete requests in bytes. This can be used by devices to inform geom of their size limitations regarding delete operations which are generally different from the read / write limits as data is not usually transferred from the host to physical device. sys/geom/geom_disk.c: - Use new d_delmaxsize to calculate the size of chunks passed through to the underlying strategy during deletes instead of using read / write optimised values. This defaults to d_maxsize if unset (0). - Moved d_maxsize default up so it can be used to default d_delmaxsize sys/cam/ata/ata_da.c: - Added d_delmaxsize calculations for TRIM and CFA sys/cam/scsi/scsi_da.c: - Added re-calculation of d_delmaxsize whenever delete_method is set. - Added kern.cam.da.X.delete_max sysctl which allows the max size for delete requests to be limited. This is useful in preventing timeouts on devices who's delete methods are slow. It should be noted that this limit is reset then the device delete method is changed and that it can only be lowered not increased from the device max. Reviewed by: mav Approved by: pjd (mentor)
2013-04-26 16:22:54 +00:00
break;
case DA_DELETE_ATA_TRIM:
sectors = (off_t)ATA_DSM_RANGE_MAX * softc->trim_max_ranges;
break;
case DA_DELETE_WS16:
Fix several potential overflows in UNMAP code. MFC after: 1 week
2015-01-26 15:47:08 +00:00
sectors = omin(softc->ws_max_blks, WS16_MAX_BLKS);
Teach GEOM and CAM about the difference between the max "size" of r/w and delete requests. sys/geom/geom_disk.h: - Added d_delmaxsize which represents the maximum size of individual device delete requests in bytes. This can be used by devices to inform geom of their size limitations regarding delete operations which are generally different from the read / write limits as data is not usually transferred from the host to physical device. sys/geom/geom_disk.c: - Use new d_delmaxsize to calculate the size of chunks passed through to the underlying strategy during deletes instead of using read / write optimised values. This defaults to d_maxsize if unset (0). - Moved d_maxsize default up so it can be used to default d_delmaxsize sys/cam/ata/ata_da.c: - Added d_delmaxsize calculations for TRIM and CFA sys/cam/scsi/scsi_da.c: - Added re-calculation of d_delmaxsize whenever delete_method is set. - Added kern.cam.da.X.delete_max sysctl which allows the max size for delete requests to be limited. This is useful in preventing timeouts on devices who's delete methods are slow. It should be noted that this limit is reset then the device delete method is changed and that it can only be lowered not increased from the device max. Reviewed by: mav Approved by: pjd (mentor)
2013-04-26 16:22:54 +00:00
break;
case DA_DELETE_ZERO:
case DA_DELETE_WS10:
Fix several potential overflows in UNMAP code. MFC after: 1 week
2015-01-26 15:47:08 +00:00
sectors = omin(softc->ws_max_blks, WS10_MAX_BLKS);
Teach GEOM and CAM about the difference between the max "size" of r/w and delete requests. sys/geom/geom_disk.h: - Added d_delmaxsize which represents the maximum size of individual device delete requests in bytes. This can be used by devices to inform geom of their size limitations regarding delete operations which are generally different from the read / write limits as data is not usually transferred from the host to physical device. sys/geom/geom_disk.c: - Use new d_delmaxsize to calculate the size of chunks passed through to the underlying strategy during deletes instead of using read / write optimised values. This defaults to d_maxsize if unset (0). - Moved d_maxsize default up so it can be used to default d_delmaxsize sys/cam/ata/ata_da.c: - Added d_delmaxsize calculations for TRIM and CFA sys/cam/scsi/scsi_da.c: - Added re-calculation of d_delmaxsize whenever delete_method is set. - Added kern.cam.da.X.delete_max sysctl which allows the max size for delete requests to be limited. This is useful in preventing timeouts on devices who's delete methods are slow. It should be noted that this limit is reset then the device delete method is changed and that it can only be lowered not increased from the device max. Reviewed by: mav Approved by: pjd (mentor)
2013-04-26 16:22:54 +00:00
break;
default:
return 0;
}
return (off_t)softc->params.secsize *
Fix several potential overflows in UNMAP code. MFC after: 1 week
2015-01-26 15:47:08 +00:00
omin(sectors, softc->params.sectors);
Teach GEOM and CAM about the difference between the max "size" of r/w and delete requests. sys/geom/geom_disk.h: - Added d_delmaxsize which represents the maximum size of individual device delete requests in bytes. This can be used by devices to inform geom of their size limitations regarding delete operations which are generally different from the read / write limits as data is not usually transferred from the host to physical device. sys/geom/geom_disk.c: - Use new d_delmaxsize to calculate the size of chunks passed through to the underlying strategy during deletes instead of using read / write optimised values. This defaults to d_maxsize if unset (0). - Moved d_maxsize default up so it can be used to default d_delmaxsize sys/cam/ata/ata_da.c: - Added d_delmaxsize calculations for TRIM and CFA sys/cam/scsi/scsi_da.c: - Added re-calculation of d_delmaxsize whenever delete_method is set. - Added kern.cam.da.X.delete_max sysctl which allows the max size for delete requests to be limited. This is useful in preventing timeouts on devices who's delete methods are slow. It should be noted that this limit is reset then the device delete method is changed and that it can only be lowered not increased from the device max. Reviewed by: mav Approved by: pjd (mentor)
2013-04-26 16:22:54 +00:00
}
Use the existence of ATA Information VPD to determine if we should attempt to query ATA functionality via ATA Pass-Through (16) as this page is defined as "must" for SATL devices, hence indicating that the device is at least likely to support Pass-Through (16). This eliminates errors produced by CTL when ATA Pass-Through (16) fails. Switch ATA probe daerror call to SF_NO_PRINT to avoid errors printing out for devices which return invalid errors. Output details about supported and choosen delete method when verbose booted. Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 14:37:23 +00:00
static void
daprobedone(struct cam_periph *periph, union ccb *ccb)
{
struct da_softc *softc;
softc = (struct da_softc *)periph->softc;
dadeletemethodchoose(softc, DA_DELETE_NONE);
Do not reread SCSI disk VPD pages on every device open. Instead of rereading VPD pages on every device open, do it only on initial device probe, and in cases when device reported via UNIT ATTENTIONs that something has changed. Capacity is still rereaded on every open because it is more critical for operation and more probable to change in run time. On my tests with Intel 530 SSDs on mps(4) HBA this change reduces time GEOM needs to retaste the device (that includes few open/close cycles) from ~150ms to ~30ms. MFC after: 2 weeks Sponsored by: iXsystems, Inc.
2014-04-30 17:38:26 +00:00
if (bootverbose && (softc->flags & DA_FLAG_ANNOUNCED) == 0) {
Use the existence of ATA Information VPD to determine if we should attempt to query ATA functionality via ATA Pass-Through (16) as this page is defined as "must" for SATL devices, hence indicating that the device is at least likely to support Pass-Through (16). This eliminates errors produced by CTL when ATA Pass-Through (16) fails. Switch ATA probe daerror call to SF_NO_PRINT to avoid errors printing out for devices which return invalid errors. Output details about supported and choosen delete method when verbose booted. Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 14:37:23 +00:00
char buf[80];
int i, sep;
snprintf(buf, sizeof(buf), "Delete methods: <");
sep = 0;
Make delete method set via kern.cam.da.X.delete_method persistent. This allows to set delete method via tunable, before device capabilities are known. Also allow ZERO method for devices not reporting LBP, if user explicitly requests it -- it may be useful if storage supports compression and WRITE SAME, but does not support UNMAP. MFC after: 2 weeks
2015-10-11 18:26:06 +00:00
for (i = 0; i <= DA_DELETE_MAX; i++) {
if ((softc->delete_available & (1 << i)) == 0 &&
i != softc->delete_method)
continue;
if (sep)
strlcat(buf, ",", sizeof(buf));
strlcat(buf, da_delete_method_names[i],
sizeof(buf));
if (i == softc->delete_method)
strlcat(buf, "(*)", sizeof(buf));
sep = 1;
Use the existence of ATA Information VPD to determine if we should attempt to query ATA functionality via ATA Pass-Through (16) as this page is defined as "must" for SATL devices, hence indicating that the device is at least likely to support Pass-Through (16). This eliminates errors produced by CTL when ATA Pass-Through (16) fails. Switch ATA probe daerror call to SF_NO_PRINT to avoid errors printing out for devices which return invalid errors. Output details about supported and choosen delete method when verbose booted. Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 14:37:23 +00:00
}
strlcat(buf, ">", sizeof(buf));
printf("%s%d: %s\n", periph->periph_name,
periph->unit_number, buf);
}
/*
* Since our peripheral may be invalidated by an error
* above or an external event, we must release our CCB
* before releasing the probe lock on the peripheral.
* The peripheral will only go away once the last lock
* is removed, and we need it around for the CCB release
* operation.
*/
xpt_release_ccb(ccb);
softc->state = DA_STATE_NORMAL;
Do not reread SCSI disk VPD pages on every device open. Instead of rereading VPD pages on every device open, do it only on initial device probe, and in cases when device reported via UNIT ATTENTIONs that something has changed. Capacity is still rereaded on every open because it is more critical for operation and more probable to change in run time. On my tests with Intel 530 SSDs on mps(4) HBA this change reduces time GEOM needs to retaste the device (that includes few open/close cycles) from ~150ms to ~30ms. MFC after: 2 weeks Sponsored by: iXsystems, Inc.
2014-04-30 17:38:26 +00:00
softc->flags |= DA_FLAG_PROBED;
Use the existence of ATA Information VPD to determine if we should attempt to query ATA functionality via ATA Pass-Through (16) as this page is defined as "must" for SATL devices, hence indicating that the device is at least likely to support Pass-Through (16). This eliminates errors produced by CTL when ATA Pass-Through (16) fails. Switch ATA probe daerror call to SF_NO_PRINT to avoid errors printing out for devices which return invalid errors. Output details about supported and choosen delete method when verbose booted. Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 14:37:23 +00:00
daschedule(periph);
wakeup(&softc->disk->d_mediasize);
Do not reread SCSI disk VPD pages on every device open. Instead of rereading VPD pages on every device open, do it only on initial device probe, and in cases when device reported via UNIT ATTENTIONs that something has changed. Capacity is still rereaded on every open because it is more critical for operation and more probable to change in run time. On my tests with Intel 530 SSDs on mps(4) HBA this change reduces time GEOM needs to retaste the device (that includes few open/close cycles) from ~150ms to ~30ms. MFC after: 2 weeks Sponsored by: iXsystems, Inc.
2014-04-30 17:38:26 +00:00
if ((softc->flags & DA_FLAG_ANNOUNCED) == 0) {
softc->flags |= DA_FLAG_ANNOUNCED;
Use the existence of ATA Information VPD to determine if we should attempt to query ATA functionality via ATA Pass-Through (16) as this page is defined as "must" for SATL devices, hence indicating that the device is at least likely to support Pass-Through (16). This eliminates errors produced by CTL when ATA Pass-Through (16) fails. Switch ATA probe daerror call to SF_NO_PRINT to avoid errors printing out for devices which return invalid errors. Output details about supported and choosen delete method when verbose booted. Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 14:37:23 +00:00
cam_periph_unhold(periph);
} else
cam_periph_release_locked(periph);
}
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
static void
dadeletemethodchoose(struct da_softc *softc, da_delete_methods default_method)
{
Make delete method set via kern.cam.da.X.delete_method persistent. This allows to set delete method via tunable, before device capabilities are known. Also allow ZERO method for devices not reporting LBP, if user explicitly requests it -- it may be useful if storage supports compression and WRITE SAME, but does not support UNMAP. MFC after: 2 weeks
2015-10-11 18:26:06 +00:00
int i, methods;
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
Make delete method set via kern.cam.da.X.delete_method persistent. This allows to set delete method via tunable, before device capabilities are known. Also allow ZERO method for devices not reporting LBP, if user explicitly requests it -- it may be useful if storage supports compression and WRITE SAME, but does not support UNMAP. MFC after: 2 weeks
2015-10-11 18:26:06 +00:00
/* If available, prefer the method requested by user. */
i = softc->delete_method_pref;
methods = softc->delete_available | (1 << DA_DELETE_DISABLE);
if (methods & (1 << i)) {
dadeletemethodset(softc, i);
return;
}
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
Make delete method set via kern.cam.da.X.delete_method persistent. This allows to set delete method via tunable, before device capabilities are known. Also allow ZERO method for devices not reporting LBP, if user explicitly requests it -- it may be useful if storage supports compression and WRITE SAME, but does not support UNMAP. MFC after: 2 weeks
2015-10-11 18:26:06 +00:00
/* Use the pre-defined order to choose the best performing delete. */
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
for (i = DA_DELETE_MIN; i <= DA_DELETE_MAX; i++) {
Make delete method set via kern.cam.da.X.delete_method persistent. This allows to set delete method via tunable, before device capabilities are known. Also allow ZERO method for devices not reporting LBP, if user explicitly requests it -- it may be useful if storage supports compression and WRITE SAME, but does not support UNMAP. MFC after: 2 weeks
2015-10-11 18:26:06 +00:00
if (i == DA_DELETE_ZERO)
continue;
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
if (softc->delete_available & (1 << i)) {
dadeletemethodset(softc, i);
return;
}
}
Make delete method set via kern.cam.da.X.delete_method persistent. This allows to set delete method via tunable, before device capabilities are known. Also allow ZERO method for devices not reporting LBP, if user explicitly requests it -- it may be useful if storage supports compression and WRITE SAME, but does not support UNMAP. MFC after: 2 weeks
2015-10-11 18:26:06 +00:00
/* Fallback to default. */
dadeletemethodset(softc, default_method);
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
}
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
static int
dadeletemethodsysctl(SYSCTL_HANDLER_ARGS)
{
char buf[16];
const char *p;
Updates delete_method sysctl changes to always maintain disk d_flags DISKFLAG_CANDELETE. While this change makes this layer consistent other layers such as UFS and ZFS BIO_DELETE support may not notice any change made manually via these device sysctls until the device is reopened via a mount. Also corrected var order in dadeletemethodsysctl PR: kern/169801 Reviewed by: pjd (mentor) Approved by: mav MFC after: 2 weeks
2013-01-10 11:57:46 +00:00
struct da_softc *softc;
Allow delete_method sysctl to be set to "DISABLE".
2014-01-07 20:12:10 +00:00
int i, error, methods, value;
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
Updates delete_method sysctl changes to always maintain disk d_flags DISKFLAG_CANDELETE. While this change makes this layer consistent other layers such as UFS and ZFS BIO_DELETE support may not notice any change made manually via these device sysctls until the device is reopened via a mount. Also corrected var order in dadeletemethodsysctl PR: kern/169801 Reviewed by: pjd (mentor) Approved by: mav MFC after: 2 weeks
2013-01-10 11:57:46 +00:00
softc = (struct da_softc *)arg1;
value = softc->delete_method;
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
if (value < 0 || value > DA_DELETE_MAX)
p = "UNKNOWN";
else
p = da_delete_method_names[value];
strncpy(buf, p, sizeof(buf));
error = sysctl_handle_string(oidp, buf, sizeof(buf), req);
if (error != 0 || req->newptr == NULL)
return (error);
Allow delete_method sysctl to be set to "DISABLE".
2014-01-07 20:12:10 +00:00
methods = softc->delete_available | (1 << DA_DELETE_DISABLE);
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
for (i = 0; i <= DA_DELETE_MAX; i++) {
Make delete method set via kern.cam.da.X.delete_method persistent. This allows to set delete method via tunable, before device capabilities are known. Also allow ZERO method for devices not reporting LBP, if user explicitly requests it -- it may be useful if storage supports compression and WRITE SAME, but does not support UNMAP. MFC after: 2 weeks
2015-10-11 18:26:06 +00:00
if (strcmp(buf, da_delete_method_names[i]) == 0)
break;
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
}
Make delete method set via kern.cam.da.X.delete_method persistent. This allows to set delete method via tunable, before device capabilities are known. Also allow ZERO method for devices not reporting LBP, if user explicitly requests it -- it may be useful if storage supports compression and WRITE SAME, but does not support UNMAP. MFC after: 2 weeks
2015-10-11 18:26:06 +00:00
if (i > DA_DELETE_MAX)
return (EINVAL);
softc->delete_method_pref = i;
dadeletemethodchoose(softc, DA_DELETE_NONE);
return (0);
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
}
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
static cam_status
daregister(struct cam_periph *periph, void *arg)
{
struct da_softc *softc;
Add a PATH_INQ flag, PIM_NO_6_BYTE, which indicates the SIM never wishes to receive 6 byte commands. Add a check for this flag to da(4) and cd(4) so that they honor it. This is a quick workaround for many devices (especially USB) that require da(4) quirks to operate. The more complete approach is to finish the new transport code which will be aware of the SCSI version a transport implements. MFC after: 1 day
2003-07-28 06:15:59 +00:00
struct ccb_pathinq cpi;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
struct ccb_getdev *cgd;
Move dynamic sysctl(8) variable creation for the cd(4) and da(4) drivers out of cdregister() and daregister(), which are run from interrupt context. The sysctl code does blocking mallocs (M_WAITOK), which causes problems if malloc(9) actually needs to sleep. The eventual fix for this issue will involve moving the CAM probe process inside a kernel thread. For now, though, I have fixed the issue by moving dynamic sysctl variable creation for these two drivers to a task queue running in a kernel thread. The existing task queues (taskqueue_swi and taskqueue_swi_giant) run in software interrupt handlers, which wouldn't fix the problem at hand. So I have created a new task queue, taskqueue_thread, that runs inside a kernel thread. (It also runs outside of Giant -- clients must explicitly acquire and release Giant in their taskqueue functions.) scsi_cd.c: Remove sysctl variable creation code from cdregister(), and move it to a new function, cdsysctlinit(). Queue cdsysctlinit() to the taskqueue_thread taskqueue once we have fully registered the cd(4) driver instance. scsi_da.c: Remove sysctl variable creation code from daregister(), and move it to move it to a new function, dasysctlinit(). Queue dasysctlinit() to the taskqueue_thread taskqueue once we have fully registered the da(4) instance. taskqueue.h: Declare the new taskqueue_thread taskqueue, update some comments. subr_taskqueue.c: Create the new kernel thread taskqueue. This taskqueue runs outside of Giant, so any functions queued to it would need to explicitly acquire/release Giant if they need it. cd.4: Update the cd(4) man page to talk about the minimum command size sysctl/loader tunable. Also note that the changer variables are available as loader tunables as well. da.4: Update the da(4) man page to cover the retry_count, default_timeout and minimum_cmd_size sysctl variables/loader tunables. Remove references to /dev/r???, they aren't used any longer. cd.9: Update the cd(9) man page to describe the CD_Q_10_BYTE_ONLY quirk. taskqueue.9: Update the taskqueue(9) man page to describe the new thread task queue, and the taskqueue_swi_giant queue. MFC after: 3 days
2003-09-03 04:46:28 +00:00
char tmpstr[80];
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
caddr_t match;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
cgd = (struct ccb_getdev *)arg;
if (cgd == NULL) {
printf("daregister: no getdev CCB, can't register device\n");
return(CAM_REQ_CMP_ERR);
}
Use malloc(9)'s M_ZERO rather than explicit bzero(9) call.
2002-10-11 09:59:22 +00:00
softc = (struct da_softc *)malloc(sizeof(*softc), M_DEVBUF,
M_NOWAIT|M_ZERO);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
if (softc == NULL) {
printf("daregister: Unable to probe new device. "
"Unable to allocate softc\n");
return(CAM_REQ_CMP_ERR);
}
LIST_INIT(&softc->pending_ccbs);
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
softc->state = DA_STATE_PROBE_RC;
Complete the bio/buf divorce for all code below devfs::strategy Exceptions: Vinum untouched. This means that it cannot be compiled. Greg Lehey is on the case. CCD not converted yet, casts to struct buf (still safe) atapi-cd casts to struct buf to examine B_PHYS
2000-04-15 05:54:02 +00:00
bioq_init(&softc->bio_queue);
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
bioq_init(&softc->delete_queue);
bioq_init(&softc->delete_run_queue);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
if (SID_IS_REMOVABLE(&cgd->inq_data))
softc->flags |= DA_FLAG_PACK_REMOVABLE;
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
softc->unmap_max_ranges = UNMAP_MAX_RANGES;
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
softc->unmap_max_lba = UNMAP_RANGE_MAX;
softc->ws_max_blks = WS16_MAX_BLKS;
softc->trim_max_ranges = ATA_TRIM_MAX_RANGES;
Adds the ability to enable / disable sorting of BIO requests queued within CAM. This can significantly improve performance particularly for SSDs which don't suffer from seek latencies. The sysctl / tunable kern.cam.sort_io_queues provides the systems default setting where:- 0 = queued BIOs are NOT sorted 1 = queued BIOs are sorted (default) Each device gets its own sysctl kern.cam.<type>.<id>.sort_io_queue Valid values are:- -1 = use system default (default) 0 = queued BIOs are NOT sorted 1 = queued BIOs are sorted Note: Additional patch will look to add automatic use of none sorted queues for none rotating media e.g. SSD's Reviewed by: scottl Approved by: pjd (mentor) MFC after: 2 weeks
2013-03-29 22:58:15 +00:00
softc->sort_io_queue = -1;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
periph->softc = softc;
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
/*
* See if this device has any quirks.
*/
match = cam_quirkmatch((caddr_t)&cgd->inq_data,
(caddr_t)da_quirk_table,
sizeof(da_quirk_table)/sizeof(*da_quirk_table),
sizeof(*da_quirk_table), scsi_inquiry_match);
if (match != NULL)
softc->quirks = ((struct da_quirk_entry *)match)->quirks;
else
softc->quirks = DA_Q_NONE;
Add a PATH_INQ flag, PIM_NO_6_BYTE, which indicates the SIM never wishes to receive 6 byte commands. Add a check for this flag to da(4) and cd(4) so that they honor it. This is a quick workaround for many devices (especially USB) that require da(4) quirks to operate. The more complete approach is to finish the new transport code which will be aware of the SCSI version a transport implements. MFC after: 1 day
2003-07-28 06:15:59 +00:00
/* Check if the SIM does not want 6 byte commands */
Separate the parallel scsi knowledge out of the core of the XPT, and modularize it so that new transports can be created. Add a transport for SATA Add a periph+protocol layer for ATA Add a driver for AHCI-compliant hardware. Add a maxio field to CAM so that drivers can advertise their max I/O capability. Modify various drivers so that they are insulated from the value of MAXPHYS. The new ATA/SATA code supports AHCI-compliant hardware, and will override the classic ATA driver if it is loaded as a module at boot time or compiled into the kernel. The stack now support NCQ (tagged queueing) for increased performance on modern SATA drives. It also supports port multipliers. ATA drives are accessed via 'ada' device nodes. ATAPI drives are accessed via 'cd' device nodes. They can all be enumerated and manipulated via camcontrol, just like SCSI drives. SCSI commands are not translated to their ATA equivalents; ATA native commands are used throughout the entire stack, including camcontrol. See the camcontrol manpage for further details. Testing this code may require that you update your fstab, and possibly modify your BIOS to enable AHCI functionality, if available. This code is very experimental at the moment. The userland ABI/API has changed, so applications will need to be recompiled. It may change further in the near future. The 'ada' device name may also change as more infrastructure is completed in this project. The goal is to eventually put all CAM busses and devices until newbus, allowing for interesting topology and management options. Few functional changes will be seen with existing SCSI/SAS/FC drivers, though the userland ABI has still changed. In the future, transports specific modules for SAS and FC may appear in order to better support the topologies and capabilities of these technologies. The modularization of CAM and the addition of the ATA/SATA modules is meant to break CAM out of the mold of being specific to SCSI, letting it grow to be a framework for arbitrary transports and protocols. It also allows drivers to be written to support discrete hardware without jeopardizing the stability of non-related hardware. While only an AHCI driver is provided now, a Silicon Image driver is also in the works. Drivers for ICH1-4, ICH5-6, PIIX, classic IDE, and any other hardware is possible and encouraged. Help with new transports is also encouraged. Submitted by: scottl, mav Approved by: re
2009-07-10 08:18:08 +00:00
bzero(&cpi, sizeof(cpi));
Replace most of priority numbers with defines. No logical changes.
2009-10-23 08:27:55 +00:00
xpt_setup_ccb(&cpi.ccb_h, periph->path, CAM_PRIORITY_NORMAL);
Add a PATH_INQ flag, PIM_NO_6_BYTE, which indicates the SIM never wishes to receive 6 byte commands. Add a check for this flag to da(4) and cd(4) so that they honor it. This is a quick workaround for many devices (especially USB) that require da(4) quirks to operate. The more complete approach is to finish the new transport code which will be aware of the SCSI version a transport implements. MFC after: 1 day
2003-07-28 06:15:59 +00:00
cpi.ccb_h.func_code = XPT_PATH_INQ;
xpt_action((union ccb *)&cpi);
if (cpi.ccb_h.status == CAM_REQ_CMP && (cpi.hba_misc & PIM_NO_6_BYTE))
softc->quirks |= DA_Q_NO_6_BYTE;
Move dynamic sysctl(8) variable creation for the cd(4) and da(4) drivers out of cdregister() and daregister(), which are run from interrupt context. The sysctl code does blocking mallocs (M_WAITOK), which causes problems if malloc(9) actually needs to sleep. The eventual fix for this issue will involve moving the CAM probe process inside a kernel thread. For now, though, I have fixed the issue by moving dynamic sysctl variable creation for these two drivers to a task queue running in a kernel thread. The existing task queues (taskqueue_swi and taskqueue_swi_giant) run in software interrupt handlers, which wouldn't fix the problem at hand. So I have created a new task queue, taskqueue_thread, that runs inside a kernel thread. (It also runs outside of Giant -- clients must explicitly acquire and release Giant in their taskqueue functions.) scsi_cd.c: Remove sysctl variable creation code from cdregister(), and move it to a new function, cdsysctlinit(). Queue cdsysctlinit() to the taskqueue_thread taskqueue once we have fully registered the cd(4) driver instance. scsi_da.c: Remove sysctl variable creation code from daregister(), and move it to move it to a new function, dasysctlinit(). Queue dasysctlinit() to the taskqueue_thread taskqueue once we have fully registered the da(4) instance. taskqueue.h: Declare the new taskqueue_thread taskqueue, update some comments. subr_taskqueue.c: Create the new kernel thread taskqueue. This taskqueue runs outside of Giant, so any functions queued to it would need to explicitly acquire/release Giant if they need it. cd.4: Update the cd(4) man page to talk about the minimum command size sysctl/loader tunable. Also note that the changer variables are available as loader tunables as well. da.4: Update the da(4) man page to cover the retry_count, default_timeout and minimum_cmd_size sysctl variables/loader tunables. Remove references to /dev/r???, they aren't used any longer. cd.9: Update the cd(9) man page to describe the CD_Q_10_BYTE_ONLY quirk. taskqueue.9: Update the taskqueue(9) man page to describe the new thread task queue, and the taskqueue_swi_giant queue. MFC after: 3 days
2003-09-03 04:46:28 +00:00
TASK_INIT(&softc->sysctl_task, 0, dasysctlinit, periph);
Fix ATAPI/USB/Firewire CDROM drive handling in cd(4) and hopefully fix a number of related problems along the way. - Automatically detect CDROM drives that can't handle 6 byte mode sense and mode select, and adjust our command size accordingly. We have to handle this in the cd(4) driver (where the buffers are allocated), since the parameter list length is different for the 6 and 10 byte mode sense commands. - Remove MODE_SENSE and MODE_SELECT translation removed in ATAPICAM and in the umass(4) driver, since there's no way for that to work properly. - Add a quirk entry for CDROM drives that just hang when they get a 6 byte mode sense or mode select. The reason for the quirk must be documented in a PR, and all quirks must be approved by ken@FreeBSD.org. This is to make sure that we fully understand why each quirk is needed. Once the CAM_NEW_TRAN_CODE is finished, we should be able to remove any such quirks, since we'll know what protocol the drive speaks (SCSI, ATAPI, etc.) and therefore whether we should use 6 or 10 byte mode sense/select commands. - Change the way the da(4) handles the no_6_byte sysctl. There is now a per-drive sysctl to set the minimum command size for that particular disk. (Since you could have multiple disks with multiple requirements in one system.) - Loader tunable support for all the sysctls in the da(4) and cd(4) drivers. - Add a CDIOCCLOSE ioctl for cd(4) (bde pointed this out a long time ago). - Add a media validation routine (cdcheckmedia()) to the cd(4) driver, to fix some problems bde pointed out a long time ago. We now allow open() to succeed no matter what, but if we don't detect valid media, the user can only issue CDIOCCLOSE or CDIOCEJECT ioctls. - The media validation routine also reads the table of contents off the drive. We use the table of contents to implement the CDIOCPLAYTRACKS ioctl using the PLAY AUDIO MSF command. The PLAY AUDIO TRACK INDEX command that we previously used was deprecated after SCSI-2. It works in every SCSI CDROM I've tried, but doesn't seem to work on ATAPI CDROM drives. We still use the play audio track index command if we don't have a valid TOC, but I suppose it'll fail anyway in that case. - Add _len() versions of scsi_mode_sense() and scsi_mode_select() so that we can specify the minimum command length. - Fix a couple of formatting problems in the sense printing code. MFC after: 4 weeks
2003-02-21 06:19:38 +00:00
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
/*
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
* Take an exclusive refcount on the periph while dastart is called
* to finish the probe. The reference will be dropped in dadone at
* the end of probe.
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
*/
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
(void)cam_periph_hold(periph, PRIBIO);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
/*
* Schedule a periodic event to occasionally send an
* ordered tag to a device.
*/
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
callout_init_mtx(&softc->sendordered_c, cam_periph_mtx(periph), 0);
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
callout_reset(&softc->sendordered_c,
Insert ordered command every 1/4 of the current command timeout, not 1/4 of the default one. Without this change setting kern.cam.ada.default_timeout to 1 instead of 30 allowed me to trigger several false positive command timeouts under heavy ZFS load on a SiI3132 siis(4) controller with 5 HDDs on a port multiplier. MFC after: 1 week
2012-02-02 19:02:15 +00:00
(da_default_timeout * hz) / DA_ORDEREDTAG_INTERVAL,
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
dasendorderedtag, softc);
MFprojects/camlock r248931: Replace some direct mutex operations with wrappers. MFC after: 2 weeks
2013-04-04 19:07:37 +00:00
cam_periph_unlock(periph);
Move TUNABLE_INT_FETCH() out of the SIM lock to avoid sleeping while holding the lock. The fix is analogous to r220618 for ada(4). Reviewed by: mav
2011-04-16 06:54:41 +00:00
/*
* RBC devices don't have to support READ(6), only READ(10).
*/
if (softc->quirks & DA_Q_NO_6_BYTE || SID_TYPE(&cgd->inq_data) == T_RBC)
softc->minimum_cmd_size = 10;
else
softc->minimum_cmd_size = 6;
/*
* Load the user's default, if any.
*/
snprintf(tmpstr, sizeof(tmpstr), "kern.cam.da.%d.minimum_cmd_size",
periph->unit_number);
TUNABLE_INT_FETCH(tmpstr, &softc->minimum_cmd_size);
/*
* 6, 10, 12 and 16 are the currently permissible values.
*/
if (softc->minimum_cmd_size < 6)
softc->minimum_cmd_size = 6;
else if ((softc->minimum_cmd_size > 6)
&& (softc->minimum_cmd_size <= 10))
softc->minimum_cmd_size = 10;
else if ((softc->minimum_cmd_size > 10)
&& (softc->minimum_cmd_size <= 12))
softc->minimum_cmd_size = 12;
else if (softc->minimum_cmd_size > 12)
softc->minimum_cmd_size = 16;
Use READ CAPACITY(16) to get information about device physical sectors. As soon as not all devices support READ CAPACITY(16), automatically fall back to READ CAPACITY(10) if CAM_REQ_INVALID or SSD_KEY_ILLEGAL_REQUEST status returned. It also provides first bits of information about Logical Block Provisioning (aka UNMAP/TRIM) support by the device.
2011-12-23 19:12:02 +00:00
/* Predict whether device may support READ CAPACITY(16). */
Add NO_RC16 quirk to make da driver avoid using READ CAPACITY(16) command if possible. Use it for Kingston JetFlash USB sticks, that are known to return garbage in response to that command.
2013-07-30 13:00:09 +00:00
if (SID_ANSI_REV(&cgd->inq_data) >= SCSI_REV_SPC3 &&
(softc->quirks & DA_Q_NO_RC16) == 0) {
Use READ CAPACITY(16) to get information about device physical sectors. As soon as not all devices support READ CAPACITY(16), automatically fall back to READ CAPACITY(10) if CAM_REQ_INVALID or SSD_KEY_ILLEGAL_REQUEST status returned. It also provides first bits of information about Logical Block Provisioning (aka UNMAP/TRIM) support by the device.
2011-12-23 19:12:02 +00:00
softc->flags |= DA_FLAG_CAN_RC16;
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
softc->state = DA_STATE_PROBE_RC16;
Use READ CAPACITY(16) to get information about device physical sectors. As soon as not all devices support READ CAPACITY(16), automatically fall back to READ CAPACITY(10) if CAM_REQ_INVALID or SSD_KEY_ILLEGAL_REQUEST status returned. It also provides first bits of information about Logical Block Provisioning (aka UNMAP/TRIM) support by the device.
2011-12-23 19:12:02 +00:00
}
Move a comment to the right place. The rearrange done in r208928 left the comment to a wrong place.
2011-04-16 08:38:11 +00:00
/*
* Register this media as a disk.
*/
Rearrange how things are done to avoid dereferencing stale pointers in the case of immediate unconfigure after configure. Hold the periph an extra count while we have the task to create sysctl context outstanding so that the periph doesn't go away unexpectedly. Sponsored by: Panasas Reviewed by: scsi@ MFC after: 1 month
2010-06-08 22:46:44 +00:00
softc->disk = disk_alloc();
Make CAM report devices with ATA/SATA transport to devstat(9) as IDE.
2011-04-14 21:25:32 +00:00
softc->disk->d_devstat = devstat_new_entry(periph->periph_name,
periph->unit_number, 0,
DEVSTAT_BS_UNAVAILABLE,
SID_TYPE(&cgd->inq_data) |
XPORT_DEVSTAT_TYPE(cpi.transport),
DEVSTAT_PRIORITY_DISK);
Rearrange how things are done to avoid dereferencing stale pointers in the case of immediate unconfigure after configure. Hold the periph an extra count while we have the task to create sysctl context outstanding so that the periph doesn't go away unexpectedly. Sponsored by: Panasas Reviewed by: scsi@ MFC after: 1 month
2010-06-08 22:46:44 +00:00
softc->disk->d_open = daopen;
softc->disk->d_close = daclose;
softc->disk->d_strategy = dastrategy;
softc->disk->d_dump = dadump;
Plumb device physical path reporting from CAM devices, through GEOM and DEVFS, and make it accessible via the diskinfo utility. Extend GEOM's generic attribute query mechanism into generic disk consumers. sys/geom/geom_disk.c: sys/geom/geom_disk.h: sys/cam/scsi/scsi_da.c: sys/cam/ata/ata_da.c: - Allow disk providers to implement a new method which can override the default BIO_GETATTR response, d_getattr(struct bio *). This function returns -1 if not handled, otherwise it returns 0 or an errno to be passed to g_io_deliver(). sys/cam/scsi/scsi_da.c: sys/cam/ata/ata_da.c: - Don't copy the serial number to dp->d_ident anymore, as the CAM XPT is now responsible for returning this information via d_getattr()->(a)dagetattr()->xpt_getatr(). sys/geom/geom_dev.c: - Implement a new ioctl, DIOCGPHYSPATH, which returns the GEOM attribute "GEOM::physpath", if possible. If the attribute request returns a zero-length string, ENOENT is returned. usr.sbin/diskinfo/diskinfo.c: - If the DIOCGPHYSPATH ioctl is successful, report physical path data when diskinfo is executed with the '-v' option. Submitted by: will Reviewed by: gibbs Sponsored by: Spectra Logic Corporation Add generic attribute change notification support to GEOM. sys/sys/geom/geom.h: Add a new attrchanged method field to both g_class and g_geom. sys/sys/geom/geom.h: sys/geom/geom_event.c: - Provide the g_attr_changed() function that providers can use to advertise attribute changes. - Perform delivery of attribute change notifications from a thread context via the standard GEOM event mechanism. sys/geom/geom_subr.c: Inherit the attrchanged method from class to geom (class instance). sys/geom/geom_disk.c: Provide disk_attr_changed() to provide g_attr_changed() access to consumers of the disk API. sys/cam/scsi/scsi_pass.c: sys/cam/scsi/scsi_da.c: sys/geom/geom_dev.c: sys/geom/geom_disk.c: Use attribute changed events to track updates to physical path information. sys/cam/scsi/scsi_da.c: Add AC_ADVINFO_CHANGED to the registered asynchronous CAM events for this driver. When this event occurs, and the updated buffer type references our physical path attribute, emit a GEOM attribute changed event via the disk_attr_changed() API. sys/cam/scsi/scsi_pass.c: Add AC_ADVINFO_CHANGED to the registered asynchronous CAM events for this driver. When this event occurs, update the physical patch devfs alias for this pass instance. Submitted by: gibbs Sponsored by: Spectra Logic Corporation
2011-06-14 17:10:32 +00:00
softc->disk->d_getattr = dagetattr;
Fix a bug which causes a panic in daopen(). The panic is caused by a da(4) instance going away while GEOM is still probing it. In this case, the GEOM disk class instance has been created by disk_create(), and the taste of the disk is queued in the GEOM event queue. While that event is queued, the da(4) instance goes away. When the open call comes into the da(4) driver, it dereferences the freed (but non-NULL) peripheral pointer provided by GEOM, which results in a panic. The solution is to add a callback to the GEOM disk code that is called when all of its resources are cleaned up. This is implemented inside GEOM by adding an optional callback that is called when all consumers have detached from a provider, and the provider is about to be deleted. scsi_cd.c, scsi_da.c: In the register routine for the cd(4) and da(4) routines, acquire a reference to the CAM peripheral instance just before we call disk_create(). Use the new GEOM disk d_gone() callback to register a callback (dadiskgonecb()/cddiskgonecb()) that decrements the peripheral reference count once GEOM has finished cleaning up its resources. In the cd(4) driver, clean up open and close behavior slightly. GEOM makes sure we only get one open() and one close call, so there is no need to set an open flag and decrement the reference count if we are not the first open. In the cd(4) driver, use cam_periph_release_locked() in a couple of error scenarios to avoid extra mutex calls. geom.h: Add a new, optional, providergone callback that is called when a provider is about to be deleted. geom_disk.h: Add a new d_gone() callback to the GEOM disk interface. Bump the DISK_VERSION to version 2. This probably should have been done after a couple of previous changes, especially the addition of the d_getattr() callback. geom_disk.c: Add a providergone callback for the disk class, g_disk_providergone(), that calls the user's d_gone() callback if it exists. Bump the DISK_VERSION to 2. geom_subr.c: In g_destroy_provider(), call the providergone callback if it has been provided. In g_new_geomf(), propagate the class's providergone callback to the new geom instance. blkfront.c: Callers of disk_create() are supposed to pass in DISK_VERSION, not an explicit disk API version number. Update the blkfront driver to do that. disk.9: Update the disk(9) man page to include information on the new d_gone() callback, as well as the previously added d_getattr() callback, d_descr field, and HBA PCI ID fields. MFC after: 5 days
2012-06-24 04:29:03 +00:00
softc->disk->d_gone = dadiskgonecb;
Rearrange how things are done to avoid dereferencing stale pointers in the case of immediate unconfigure after configure. Hold the periph an extra count while we have the task to create sysctl context outstanding so that the periph doesn't go away unexpectedly. Sponsored by: Panasas Reviewed by: scsi@ MFC after: 1 month
2010-06-08 22:46:44 +00:00
softc->disk->d_name = "da";
softc->disk->d_drv1 = periph;
if (cpi.maxio == 0)
Respect MAXIMUM TRANSFER LENGTH field of Block Limits VPD page. Nobody yet reported disk supporting I/Os less then our MAXPHYS value, but since we any way have code to read Block Limits VPD page, that is easy. MFC after: 2 weeks
2014-04-30 19:44:31 +00:00
softc->maxio = DFLTPHYS; /* traditional default */
Rearrange how things are done to avoid dereferencing stale pointers in the case of immediate unconfigure after configure. Hold the periph an extra count while we have the task to create sysctl context outstanding so that the periph doesn't go away unexpectedly. Sponsored by: Panasas Reviewed by: scsi@ MFC after: 1 month
2010-06-08 22:46:44 +00:00
else if (cpi.maxio > MAXPHYS)
Respect MAXIMUM TRANSFER LENGTH field of Block Limits VPD page. Nobody yet reported disk supporting I/Os less then our MAXPHYS value, but since we any way have code to read Block Limits VPD page, that is easy. MFC after: 2 weeks
2014-04-30 19:44:31 +00:00
softc->maxio = MAXPHYS; /* for safety */
Rearrange how things are done to avoid dereferencing stale pointers in the case of immediate unconfigure after configure. Hold the periph an extra count while we have the task to create sysctl context outstanding so that the periph doesn't go away unexpectedly. Sponsored by: Panasas Reviewed by: scsi@ MFC after: 1 month
2010-06-08 22:46:44 +00:00
else
Respect MAXIMUM TRANSFER LENGTH field of Block Limits VPD page. Nobody yet reported disk supporting I/Os less then our MAXPHYS value, but since we any way have code to read Block Limits VPD page, that is easy. MFC after: 2 weeks
2014-04-30 19:44:31 +00:00
softc->maxio = cpi.maxio;
softc->disk->d_maxsize = softc->maxio;
Rearrange how things are done to avoid dereferencing stale pointers in the case of immediate unconfigure after configure. Hold the periph an extra count while we have the task to create sysctl context outstanding so that the periph doesn't go away unexpectedly. Sponsored by: Panasas Reviewed by: scsi@ MFC after: 1 month
2010-06-08 22:46:44 +00:00
softc->disk->d_unit = periph->unit_number;
Merge GEOM direct dispatch changes from the projects/camlock branch. When safety requirements are met, it allows to avoid passing I/O requests to GEOM g_up/g_down thread, executing them directly in the caller context. That allows to avoid CPU bottlenecks in g_up/g_down threads, plus avoid several context switches per I/O. The defined now safety requirements are: - caller should not hold any locks and should be reenterable; - callee should not depend on GEOM dual-threaded concurency semantics; - on the way down, if request is unmapped while callee doesn't support it, the context should be sleepable; - kernel thread stack usage should be below 50%. To keep compatibility with GEOM classes not meeting above requirements new provider and consumer flags added: - G_CF_DIRECT_SEND -- consumer code meets caller requirements (request); - G_CF_DIRECT_RECEIVE -- consumer code meets callee requirements (done); - G_PF_DIRECT_SEND -- provider code meets caller requirements (done); - G_PF_DIRECT_RECEIVE -- provider code meets callee requirements (request). Capable GEOM class can set them, allowing direct dispatch in cases where it is safe. If any of requirements are not met, request is queued to g_up or g_down thread same as before. Such GEOM classes were reviewed and updated to support direct dispatch: CONCAT, DEV, DISK, GATE, MD, MIRROR, MULTIPATH, NOP, PART, RAID, STRIPE, VFS, ZERO, ZFS::VDEV, ZFS::ZVOL, all classes based on g_slice KPI (LABEL, MAP, FLASHMAP, etc). To declare direct completion capability disk(9) KPI got new flag equivalent to G_PF_DIRECT_SEND -- DISKFLAG_DIRECT_COMPLETION. da(4) and ada(4) disk drivers got it set now thanks to earlier CAM locking work. This change more then twice increases peak block storage performance on systems with manu CPUs, together with earlier CAM locking changes reaching more then 1 million IOPS (512 byte raw reads from 16 SATA SSDs on 4 HBAs to 256 user-level threads). Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-22 08:22:19 +00:00
softc->disk->d_flags = DISKFLAG_DIRECT_COMPLETION;
Rearrange how things are done to avoid dereferencing stale pointers in the case of immediate unconfigure after configure. Hold the periph an extra count while we have the task to create sysctl context outstanding so that the periph doesn't go away unexpectedly. Sponsored by: Panasas Reviewed by: scsi@ MFC after: 1 month
2010-06-08 22:46:44 +00:00
if ((softc->quirks & DA_Q_NO_SYNC_CACHE) == 0)
softc->disk->d_flags |= DISKFLAG_CANFLUSHCACHE;
Support unmapped i/o for the md(4). The vnode-backed md(4) has to map the unmapped bio because VOP_READ() and VOP_WRITE() interfaces do not allow to pass unmapped requests to the filesystem. Vnode-backed md(4) uses pbufs instead of relying on the bio_transient_map, to avoid usual md deadlock. Sponsored by: The FreeBSD Foundation Tested by: pho, scottl
2013-03-19 15:01:50 +00:00
if ((cpi.hba_misc & PIM_UNMAPPED) != 0)
softc->disk->d_flags |= DISKFLAG_UNMAPPED_BIO;
Missed a file in r219056: add disk description for da(4).
2011-02-26 23:30:32 +00:00
cam_strvis(softc->disk->d_descr, cgd->inq_data.vendor,
sizeof(cgd->inq_data.vendor), sizeof(softc->disk->d_descr));
strlcat(softc->disk->d_descr, " ", sizeof(softc->disk->d_descr));
cam_strvis(&softc->disk->d_descr[strlen(softc->disk->d_descr)],
cgd->inq_data.product, sizeof(cgd->inq_data.product),
sizeof(softc->disk->d_descr) - strlen(softc->disk->d_descr));
Export PCI IDs of ATA/SATA controllers through CAM and ata(4) layers to GEOM. This information needed for proper soft-RAID's on-disk metadata reading and writing.
2010-07-25 15:43:52 +00:00
softc->disk->d_hba_vendor = cpi.hba_vendor;
softc->disk->d_hba_device = cpi.hba_device;
softc->disk->d_hba_subvendor = cpi.hba_subvendor;
softc->disk->d_hba_subdevice = cpi.hba_subdevice;
Fix a bug which causes a panic in daopen(). The panic is caused by a da(4) instance going away while GEOM is still probing it. In this case, the GEOM disk class instance has been created by disk_create(), and the taste of the disk is queued in the GEOM event queue. While that event is queued, the da(4) instance goes away. When the open call comes into the da(4) driver, it dereferences the freed (but non-NULL) peripheral pointer provided by GEOM, which results in a panic. The solution is to add a callback to the GEOM disk code that is called when all of its resources are cleaned up. This is implemented inside GEOM by adding an optional callback that is called when all consumers have detached from a provider, and the provider is about to be deleted. scsi_cd.c, scsi_da.c: In the register routine for the cd(4) and da(4) routines, acquire a reference to the CAM peripheral instance just before we call disk_create(). Use the new GEOM disk d_gone() callback to register a callback (dadiskgonecb()/cddiskgonecb()) that decrements the peripheral reference count once GEOM has finished cleaning up its resources. In the cd(4) driver, clean up open and close behavior slightly. GEOM makes sure we only get one open() and one close call, so there is no need to set an open flag and decrement the reference count if we are not the first open. In the cd(4) driver, use cam_periph_release_locked() in a couple of error scenarios to avoid extra mutex calls. geom.h: Add a new, optional, providergone callback that is called when a provider is about to be deleted. geom_disk.h: Add a new d_gone() callback to the GEOM disk interface. Bump the DISK_VERSION to version 2. This probably should have been done after a couple of previous changes, especially the addition of the d_getattr() callback. geom_disk.c: Add a providergone callback for the disk class, g_disk_providergone(), that calls the user's d_gone() callback if it exists. Bump the DISK_VERSION to 2. geom_subr.c: In g_destroy_provider(), call the providergone callback if it has been provided. In g_new_geomf(), propagate the class's providergone callback to the new geom instance. blkfront.c: Callers of disk_create() are supposed to pass in DISK_VERSION, not an explicit disk API version number. Update the blkfront driver to do that. disk.9: Update the disk(9) man page to include information on the new d_gone() callback, as well as the previously added d_getattr() callback, d_descr field, and HBA PCI ID fields. MFC after: 5 days
2012-06-24 04:29:03 +00:00
/*
* Acquire a reference to the periph before we register with GEOM.
* We'll release this reference once GEOM calls us back (via
* dadiskgonecb()) telling us that our provider has been freed.
*/
if (cam_periph_acquire(periph) != CAM_REQ_CMP) {
xpt_print(periph->path, "%s: lost periph during "
"registration!\n", __func__);
MFprojects/camlock r248931: Replace some direct mutex operations with wrappers. MFC after: 2 weeks
2013-04-04 19:07:37 +00:00
cam_periph_lock(periph);
Fix a bug which causes a panic in daopen(). The panic is caused by a da(4) instance going away while GEOM is still probing it. In this case, the GEOM disk class instance has been created by disk_create(), and the taste of the disk is queued in the GEOM event queue. While that event is queued, the da(4) instance goes away. When the open call comes into the da(4) driver, it dereferences the freed (but non-NULL) peripheral pointer provided by GEOM, which results in a panic. The solution is to add a callback to the GEOM disk code that is called when all of its resources are cleaned up. This is implemented inside GEOM by adding an optional callback that is called when all consumers have detached from a provider, and the provider is about to be deleted. scsi_cd.c, scsi_da.c: In the register routine for the cd(4) and da(4) routines, acquire a reference to the CAM peripheral instance just before we call disk_create(). Use the new GEOM disk d_gone() callback to register a callback (dadiskgonecb()/cddiskgonecb()) that decrements the peripheral reference count once GEOM has finished cleaning up its resources. In the cd(4) driver, clean up open and close behavior slightly. GEOM makes sure we only get one open() and one close call, so there is no need to set an open flag and decrement the reference count if we are not the first open. In the cd(4) driver, use cam_periph_release_locked() in a couple of error scenarios to avoid extra mutex calls. geom.h: Add a new, optional, providergone callback that is called when a provider is about to be deleted. geom_disk.h: Add a new d_gone() callback to the GEOM disk interface. Bump the DISK_VERSION to version 2. This probably should have been done after a couple of previous changes, especially the addition of the d_getattr() callback. geom_disk.c: Add a providergone callback for the disk class, g_disk_providergone(), that calls the user's d_gone() callback if it exists. Bump the DISK_VERSION to 2. geom_subr.c: In g_destroy_provider(), call the providergone callback if it has been provided. In g_new_geomf(), propagate the class's providergone callback to the new geom instance. blkfront.c: Callers of disk_create() are supposed to pass in DISK_VERSION, not an explicit disk API version number. Update the blkfront driver to do that. disk.9: Update the disk(9) man page to include information on the new d_gone() callback, as well as the previously added d_getattr() callback, d_descr field, and HBA PCI ID fields. MFC after: 5 days
2012-06-24 04:29:03 +00:00
return (CAM_REQ_CMP_ERR);
}
Rearrange how things are done to avoid dereferencing stale pointers in the case of immediate unconfigure after configure. Hold the periph an extra count while we have the task to create sysctl context outstanding so that the periph doesn't go away unexpectedly. Sponsored by: Panasas Reviewed by: scsi@ MFC after: 1 month
2010-06-08 22:46:44 +00:00
disk_create(softc->disk, DISK_VERSION);
MFprojects/camlock r248931: Replace some direct mutex operations with wrappers. MFC after: 2 weeks
2013-04-04 19:07:37 +00:00
cam_periph_lock(periph);
Rearrange how things are done to avoid dereferencing stale pointers in the case of immediate unconfigure after configure. Hold the periph an extra count while we have the task to create sysctl context outstanding so that the periph doesn't go away unexpectedly. Sponsored by: Panasas Reviewed by: scsi@ MFC after: 1 month
2010-06-08 22:46:44 +00:00
Plumb device physical path reporting from CAM devices, through GEOM and DEVFS, and make it accessible via the diskinfo utility. Extend GEOM's generic attribute query mechanism into generic disk consumers. sys/geom/geom_disk.c: sys/geom/geom_disk.h: sys/cam/scsi/scsi_da.c: sys/cam/ata/ata_da.c: - Allow disk providers to implement a new method which can override the default BIO_GETATTR response, d_getattr(struct bio *). This function returns -1 if not handled, otherwise it returns 0 or an errno to be passed to g_io_deliver(). sys/cam/scsi/scsi_da.c: sys/cam/ata/ata_da.c: - Don't copy the serial number to dp->d_ident anymore, as the CAM XPT is now responsible for returning this information via d_getattr()->(a)dagetattr()->xpt_getatr(). sys/geom/geom_dev.c: - Implement a new ioctl, DIOCGPHYSPATH, which returns the GEOM attribute "GEOM::physpath", if possible. If the attribute request returns a zero-length string, ENOENT is returned. usr.sbin/diskinfo/diskinfo.c: - If the DIOCGPHYSPATH ioctl is successful, report physical path data when diskinfo is executed with the '-v' option. Submitted by: will Reviewed by: gibbs Sponsored by: Spectra Logic Corporation Add generic attribute change notification support to GEOM. sys/sys/geom/geom.h: Add a new attrchanged method field to both g_class and g_geom. sys/sys/geom/geom.h: sys/geom/geom_event.c: - Provide the g_attr_changed() function that providers can use to advertise attribute changes. - Perform delivery of attribute change notifications from a thread context via the standard GEOM event mechanism. sys/geom/geom_subr.c: Inherit the attrchanged method from class to geom (class instance). sys/geom/geom_disk.c: Provide disk_attr_changed() to provide g_attr_changed() access to consumers of the disk API. sys/cam/scsi/scsi_pass.c: sys/cam/scsi/scsi_da.c: sys/geom/geom_dev.c: sys/geom/geom_disk.c: Use attribute changed events to track updates to physical path information. sys/cam/scsi/scsi_da.c: Add AC_ADVINFO_CHANGED to the registered asynchronous CAM events for this driver. When this event occurs, and the updated buffer type references our physical path attribute, emit a GEOM attribute changed event via the disk_attr_changed() API. sys/cam/scsi/scsi_pass.c: Add AC_ADVINFO_CHANGED to the registered asynchronous CAM events for this driver. When this event occurs, update the physical patch devfs alias for this pass instance. Submitted by: gibbs Sponsored by: Spectra Logic Corporation
2011-06-14 17:10:32 +00:00
/*
* Add async callbacks for events of interest.
* I don't bother checking if this fails as,
* in most cases, the system will function just
* fine without them and the only alternative
* would be to not attach the device on failure.
*/
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
xpt_register_async(AC_SENT_BDR | AC_BUS_RESET | AC_LOST_DEVICE |
AC_ADVINFO_CHANGED | AC_SCSI_AEN | AC_UNIT_ATTENTION,
daasync, periph, periph->path);
Plumb device physical path reporting from CAM devices, through GEOM and DEVFS, and make it accessible via the diskinfo utility. Extend GEOM's generic attribute query mechanism into generic disk consumers. sys/geom/geom_disk.c: sys/geom/geom_disk.h: sys/cam/scsi/scsi_da.c: sys/cam/ata/ata_da.c: - Allow disk providers to implement a new method which can override the default BIO_GETATTR response, d_getattr(struct bio *). This function returns -1 if not handled, otherwise it returns 0 or an errno to be passed to g_io_deliver(). sys/cam/scsi/scsi_da.c: sys/cam/ata/ata_da.c: - Don't copy the serial number to dp->d_ident anymore, as the CAM XPT is now responsible for returning this information via d_getattr()->(a)dagetattr()->xpt_getatr(). sys/geom/geom_dev.c: - Implement a new ioctl, DIOCGPHYSPATH, which returns the GEOM attribute "GEOM::physpath", if possible. If the attribute request returns a zero-length string, ENOENT is returned. usr.sbin/diskinfo/diskinfo.c: - If the DIOCGPHYSPATH ioctl is successful, report physical path data when diskinfo is executed with the '-v' option. Submitted by: will Reviewed by: gibbs Sponsored by: Spectra Logic Corporation Add generic attribute change notification support to GEOM. sys/sys/geom/geom.h: Add a new attrchanged method field to both g_class and g_geom. sys/sys/geom/geom.h: sys/geom/geom_event.c: - Provide the g_attr_changed() function that providers can use to advertise attribute changes. - Perform delivery of attribute change notifications from a thread context via the standard GEOM event mechanism. sys/geom/geom_subr.c: Inherit the attrchanged method from class to geom (class instance). sys/geom/geom_disk.c: Provide disk_attr_changed() to provide g_attr_changed() access to consumers of the disk API. sys/cam/scsi/scsi_pass.c: sys/cam/scsi/scsi_da.c: sys/geom/geom_dev.c: sys/geom/geom_disk.c: Use attribute changed events to track updates to physical path information. sys/cam/scsi/scsi_da.c: Add AC_ADVINFO_CHANGED to the registered asynchronous CAM events for this driver. When this event occurs, and the updated buffer type references our physical path attribute, emit a GEOM attribute changed event via the disk_attr_changed() API. sys/cam/scsi/scsi_pass.c: Add AC_ADVINFO_CHANGED to the registered asynchronous CAM events for this driver. When this event occurs, update the physical patch devfs alias for this pass instance. Submitted by: gibbs Sponsored by: Spectra Logic Corporation
2011-06-14 17:10:32 +00:00
/*
* Emit an attribute changed notification just in case
* physical path information arrived before our async
* event handler was registered, but after anyone attaching
* to our disk device polled it.
*/
disk_attr_changed(softc->disk, "GEOM::physpath", M_NOWAIT);
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
/*
* Schedule a periodic media polling events.
*/
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
callout_init_mtx(&softc->mediapoll_c, cam_periph_mtx(periph), 0);
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
if ((softc->flags & DA_FLAG_PACK_REMOVABLE) &&
(cgd->inq_flags & SID_AEN) == 0 &&
da_poll_period != 0)
callout_reset(&softc->mediapoll_c, da_poll_period * hz,
damediapoll, periph);
Rearrange how things are done to avoid dereferencing stale pointers in the case of immediate unconfigure after configure. Hold the periph an extra count while we have the task to create sysctl context outstanding so that the periph doesn't go away unexpectedly. Sponsored by: Panasas Reviewed by: scsi@ MFC after: 1 month
2010-06-08 22:46:44 +00:00
xpt_schedule(periph, CAM_PRIORITY_DEV);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
return(CAM_REQ_CMP);
}
static void
dastart(struct cam_periph *periph, union ccb *start_ccb)
{
struct da_softc *softc;
softc = (struct da_softc *)periph->softc;
Tune and add some missing CAM_DEBUG() points for better consistency.
2012-06-05 09:45:42 +00:00
CAM_DEBUG(periph->path, CAM_DEBUG_TRACE, ("dastart\n"));
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
skipstate:
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
switch (softc->state) {
case DA_STATE_NORMAL:
{
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
struct bio *bp;
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
uint8_t tag_code;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
/* Run BIO_DELETE if not running yet. */
if (!softc->delete_running &&
(bp = bioq_first(&softc->delete_queue)) != NULL) {
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
if (softc->delete_func != NULL) {
softc->delete_func(periph, start_ccb, bp);
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
goto out;
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
} else {
bioq_flush(&softc->delete_queue, NULL, 0);
/* FALLTHROUGH */
}
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
}
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
/* Run regular command. */
bp = bioq_takefirst(&softc->bio_queue);
if (bp == NULL) {
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
if (softc->tur) {
softc->tur = 0;
scsi_test_unit_ready(&start_ccb->csio,
/*retries*/ da_retry_count,
dadone,
MSG_SIMPLE_Q_TAG,
SSD_FULL_SIZE,
da_default_timeout * 1000);
start_ccb->ccb_h.ccb_bp = NULL;
start_ccb->ccb_h.ccb_state = DA_CCB_TUR;
xpt_action(start_ccb);
} else
xpt_release_ccb(start_ccb);
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
break;
}
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
if (softc->tur) {
softc->tur = 0;
cam_periph_release_locked(periph);
}
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
if ((bp->bio_flags & BIO_ORDERED) != 0 ||
(softc->flags & DA_FLAG_NEED_OTAG) != 0) {
softc->flags &= ~DA_FLAG_NEED_OTAG;
Some microoptimizations for da and ada drivers: - Replace ordered_tag_count counter with single flag; - From da remove outstanding_cmds counter, duplicating pending_ccbs list; - From da_softc remove unused links field.
2013-10-24 14:05:44 +00:00
softc->flags |= DA_FLAG_WAS_OTAG;
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
tag_code = MSG_ORDERED_Q_TAG;
} else {
tag_code = MSG_SIMPLE_Q_TAG;
}
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
switch (bp->bio_cmd) {
case BIO_WRITE:
Synchronize device cache on close only if there were some write operations. While these operations are not really needed otherwise, at least for SCSI they may cause extra errors if some other initiator holds write exclusive reservation on the LUN (SYNCHRONIZE CACHE handled as "write" operation).
2013-07-27 22:44:55 +00:00
case BIO_READ:
Add asynchronous command support to the pass(4) driver, and the new camdd(8) utility. CCBs may be queued to the driver via the new CAMIOQUEUE ioctl, and completed CCBs may be retrieved via the CAMIOGET ioctl. User processes can use poll(2) or kevent(2) to get notification when I/O has completed. While the existing CAMIOCOMMAND blocking ioctl interface only supports user virtual data pointers in a CCB (generally only one per CCB), the new CAMIOQUEUE ioctl supports user virtual and physical address pointers, as well as user virtual and physical scatter/gather lists. This allows user applications to have more flexibility in their data handling operations. Kernel memory for data transferred via the queued interface is allocated from the zone allocator in MAXPHYS sized chunks, and user data is copied in and out. This is likely faster than the vmapbuf()/vunmapbuf() method used by the CAMIOCOMMAND ioctl in configurations with many processors (there are more TLB shootdowns caused by the mapping/unmapping operation) but may not be as fast as running with unmapped I/O. The new memory handling model for user requests also allows applications to send CCBs with request sizes that are larger than MAXPHYS. The pass(4) driver now limits queued requests to the I/O size listed by the SIM driver in the maxio field in the Path Inquiry (XPT_PATH_INQ) CCB. There are some things things would be good to add: 1. Come up with a way to do unmapped I/O on multiple buffers. Currently the unmapped I/O interface operates on a struct bio, which includes only one address and length. It would be nice to be able to send an unmapped scatter/gather list down to busdma. This would allow eliminating the copy we currently do for data. 2. Add an ioctl to list currently outstanding CCBs in the various queues. 3. Add an ioctl to cancel a request, or use the XPT_ABORT CCB to do that. 4. Test physical address support. Virtual pointers and scatter gather lists have been tested, but I have not yet tested physical addresses or scatter/gather lists. 5. Investigate multiple queue support. At the moment there is one queue of commands per pass(4) device. If multiple processes open the device, they will submit I/O into the same queue and get events for the same completions. This is probably the right model for most applications, but it is something that could be changed later on. Also, add a new utility, camdd(8) that uses the asynchronous pass(4) driver interface. This utility is intended to be a basic data transfer/copy utility, a simple benchmark utility, and an example of how to use the asynchronous pass(4) interface. It can copy data to and from pass(4) devices using any target queue depth, starting offset and blocksize for the input and ouptut devices. It currently only supports SCSI devices, but could be easily extended to support ATA devices. It can also copy data to and from regular files, block devices, tape devices, pipes, stdin, and stdout. It does not support queueing multiple commands to any of those targets, since it uses the standard read(2)/write(2)/writev(2)/readv(2) system calls. The I/O is done by two threads, one for the reader and one for the writer. The reader thread sends completed read requests to the writer thread in strictly sequential order, even if they complete out of order. That could be modified later on for random I/O patterns or slightly out of order I/O. camdd(8) uses kqueue(2)/kevent(2) to get I/O completion events from the pass(4) driver and also to send request notifications internally. For pass(4) devcies, camdd(8) uses a single buffer (CAM_DATA_VADDR) per CAM CCB on the reading side, and a scatter/gather list (CAM_DATA_SG) on the writing side. In addition to testing both interfaces, this makes any potential reblocking of I/O easier. No data is copied between the reader and the writer, but rather the reader's buffers are split into multiple I/O requests or combined into a single I/O request depending on the input and output blocksize. For the file I/O path, camdd(8) also uses a single buffer (read(2), write(2), pread(2) or pwrite(2)) on reads, and a scatter/gather list (readv(2), writev(2), preadv(2), pwritev(2)) on writes. Things that would be nice to do for camdd(8) eventually: 1. Add support for I/O pattern generation. Patterns like all zeros, all ones, LBA-based patterns, random patterns, etc. Right Now you can always use /dev/zero, /dev/random, etc. 2. Add support for a "sink" mode, so we do only reads with no writes. Right now, you can use /dev/null. 3. Add support for automatic queue depth probing, so that we can figure out the right queue depth on the input and output side for maximum throughput. At the moment it defaults to 6. 4. Add support for SATA device passthrough I/O. 5. Add support for random LBAs and/or lengths on the input and output sides. 6. Track average per-I/O latency and busy time. The busy time and latency could also feed in to the automatic queue depth determination. sys/cam/scsi/scsi_pass.h: Define two new ioctls, CAMIOQUEUE and CAMIOGET, that queue and fetch asynchronous CAM CCBs respectively. Although these ioctls do not have a declared argument, they both take a union ccb pointer. If we declare a size here, the ioctl code in sys/kern/sys_generic.c will malloc and free a buffer for either the CCB or the CCB pointer (depending on how it is declared). Since we have to keep a copy of the CCB (which is fairly large) anyway, having the ioctl malloc and free a CCB for each call is wasteful. sys/cam/scsi/scsi_pass.c: Add asynchronous CCB support. Add two new ioctls, CAMIOQUEUE and CAMIOGET. CAMIOQUEUE adds a CCB to the incoming queue. The CCB is executed immediately (and moved to the active queue) if it is an immediate CCB, but otherwise it will be executed in passstart() when a CCB is available from the transport layer. When CCBs are completed (because they are immediate or passdone() if they are queued), they are put on the done queue. If we get the final close on the device before all pending I/O is complete, all active I/O is moved to the abandoned queue and we increment the peripheral reference count so that the peripheral driver instance doesn't go away before all pending I/O is done. The new passcreatezone() function is called on the first call to the CAMIOQUEUE ioctl on a given device to allocate the UMA zones for I/O requests and S/G list buffers. This may be good to move off to a taskqueue at some point. The new passmemsetup() function allocates memory and scatter/gather lists to hold the user's data, and copies in any data that needs to be written. For virtual pointers (CAM_DATA_VADDR), the kernel buffer is malloced from the new pass(4) driver malloc bucket. For virtual scatter/gather lists (CAM_DATA_SG), buffers are allocated from a new per-pass(9) UMA zone in MAXPHYS-sized chunks. Physical pointers are passed in unchanged. We have support for up to 16 scatter/gather segments (for the user and kernel S/G lists) in the default struct pass_io_req, so requests with longer S/G lists require an extra kernel malloc. The new passcopysglist() function copies a user scatter/gather list to a kernel scatter/gather list. The number of elements in each list may be different, but (obviously) the amount of data stored has to be identical. The new passmemdone() function copies data out for the CAM_DATA_VADDR and CAM_DATA_SG cases. The new passiocleanup() function restores data pointers in user CCBs and frees memory. Add new functions to support kqueue(2)/kevent(2): passreadfilt() tells kevent whether or not the done queue is empty. passkqfilter() adds a knote to our list. passreadfiltdetach() removes a knote from our list. Add a new function, passpoll(), for poll(2)/select(2) to use. Add devstat(9) support for the queued CCB path. sys/cam/ata/ata_da.c: Add support for the BIO_VLIST bio type. sys/cam/cam_ccb.h: Add a new enumeration for the xflags field in the CCB header. (This doesn't change the CCB header, just adds an enumeration to use.) sys/cam/cam_xpt.c: Add a new function, xpt_setup_ccb_flags(), that allows specifying CCB flags. sys/cam/cam_xpt.h: Add a prototype for xpt_setup_ccb_flags(). sys/cam/scsi/scsi_da.c: Add support for BIO_VLIST. sys/dev/md/md.c: Add BIO_VLIST support to md(4). sys/geom/geom_disk.c: Add BIO_VLIST support to the GEOM disk class. Re-factor the I/O size limiting code in g_disk_start() a bit. sys/kern/subr_bus_dma.c: Change _bus_dmamap_load_vlist() to take a starting offset and length. Add a new function, _bus_dmamap_load_pages(), that will load a list of physical pages starting at an offset. Update _bus_dmamap_load_bio() to allow loading BIO_VLIST bios. Allow unmapped I/O to start at an offset. sys/kern/subr_uio.c: Add two new functions, physcopyin_vlist() and physcopyout_vlist(). sys/pc98/include/bus.h: Guard kernel-only parts of the pc98 machine/bus.h header with #ifdef _KERNEL. This allows userland programs to include <machine/bus.h> to get the definition of bus_addr_t and bus_size_t. sys/sys/bio.h: Add a new bio flag, BIO_VLIST. sys/sys/uio.h: Add prototypes for physcopyin_vlist() and physcopyout_vlist(). share/man/man4/pass.4: Document the CAMIOQUEUE and CAMIOGET ioctls. usr.sbin/Makefile: Add camdd. usr.sbin/camdd/Makefile: Add a makefile for camdd(8). usr.sbin/camdd/camdd.8: Man page for camdd(8). usr.sbin/camdd/camdd.c: The new camdd(8) utility. Sponsored by: Spectra Logic MFC after: 1 week
2015-12-03 20:54:55 +00:00
{
void *data_ptr;
int rw_op;
if (bp->bio_cmd == BIO_WRITE) {
softc->flags |= DA_FLAG_DIRTY;
rw_op = SCSI_RW_WRITE;
} else {
rw_op = SCSI_RW_READ;
}
data_ptr = bp->bio_data;
if ((bp->bio_flags & (BIO_UNMAPPED|BIO_VLIST)) != 0) {
rw_op |= SCSI_RW_BIO;
data_ptr = bp;
}
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
scsi_read_write(&start_ccb->csio,
/*retries*/da_retry_count,
/*cbfcnp*/dadone,
/*tag_action*/tag_code,
Add asynchronous command support to the pass(4) driver, and the new camdd(8) utility. CCBs may be queued to the driver via the new CAMIOQUEUE ioctl, and completed CCBs may be retrieved via the CAMIOGET ioctl. User processes can use poll(2) or kevent(2) to get notification when I/O has completed. While the existing CAMIOCOMMAND blocking ioctl interface only supports user virtual data pointers in a CCB (generally only one per CCB), the new CAMIOQUEUE ioctl supports user virtual and physical address pointers, as well as user virtual and physical scatter/gather lists. This allows user applications to have more flexibility in their data handling operations. Kernel memory for data transferred via the queued interface is allocated from the zone allocator in MAXPHYS sized chunks, and user data is copied in and out. This is likely faster than the vmapbuf()/vunmapbuf() method used by the CAMIOCOMMAND ioctl in configurations with many processors (there are more TLB shootdowns caused by the mapping/unmapping operation) but may not be as fast as running with unmapped I/O. The new memory handling model for user requests also allows applications to send CCBs with request sizes that are larger than MAXPHYS. The pass(4) driver now limits queued requests to the I/O size listed by the SIM driver in the maxio field in the Path Inquiry (XPT_PATH_INQ) CCB. There are some things things would be good to add: 1. Come up with a way to do unmapped I/O on multiple buffers. Currently the unmapped I/O interface operates on a struct bio, which includes only one address and length. It would be nice to be able to send an unmapped scatter/gather list down to busdma. This would allow eliminating the copy we currently do for data. 2. Add an ioctl to list currently outstanding CCBs in the various queues. 3. Add an ioctl to cancel a request, or use the XPT_ABORT CCB to do that. 4. Test physical address support. Virtual pointers and scatter gather lists have been tested, but I have not yet tested physical addresses or scatter/gather lists. 5. Investigate multiple queue support. At the moment there is one queue of commands per pass(4) device. If multiple processes open the device, they will submit I/O into the same queue and get events for the same completions. This is probably the right model for most applications, but it is something that could be changed later on. Also, add a new utility, camdd(8) that uses the asynchronous pass(4) driver interface. This utility is intended to be a basic data transfer/copy utility, a simple benchmark utility, and an example of how to use the asynchronous pass(4) interface. It can copy data to and from pass(4) devices using any target queue depth, starting offset and blocksize for the input and ouptut devices. It currently only supports SCSI devices, but could be easily extended to support ATA devices. It can also copy data to and from regular files, block devices, tape devices, pipes, stdin, and stdout. It does not support queueing multiple commands to any of those targets, since it uses the standard read(2)/write(2)/writev(2)/readv(2) system calls. The I/O is done by two threads, one for the reader and one for the writer. The reader thread sends completed read requests to the writer thread in strictly sequential order, even if they complete out of order. That could be modified later on for random I/O patterns or slightly out of order I/O. camdd(8) uses kqueue(2)/kevent(2) to get I/O completion events from the pass(4) driver and also to send request notifications internally. For pass(4) devcies, camdd(8) uses a single buffer (CAM_DATA_VADDR) per CAM CCB on the reading side, and a scatter/gather list (CAM_DATA_SG) on the writing side. In addition to testing both interfaces, this makes any potential reblocking of I/O easier. No data is copied between the reader and the writer, but rather the reader's buffers are split into multiple I/O requests or combined into a single I/O request depending on the input and output blocksize. For the file I/O path, camdd(8) also uses a single buffer (read(2), write(2), pread(2) or pwrite(2)) on reads, and a scatter/gather list (readv(2), writev(2), preadv(2), pwritev(2)) on writes. Things that would be nice to do for camdd(8) eventually: 1. Add support for I/O pattern generation. Patterns like all zeros, all ones, LBA-based patterns, random patterns, etc. Right Now you can always use /dev/zero, /dev/random, etc. 2. Add support for a "sink" mode, so we do only reads with no writes. Right now, you can use /dev/null. 3. Add support for automatic queue depth probing, so that we can figure out the right queue depth on the input and output side for maximum throughput. At the moment it defaults to 6. 4. Add support for SATA device passthrough I/O. 5. Add support for random LBAs and/or lengths on the input and output sides. 6. Track average per-I/O latency and busy time. The busy time and latency could also feed in to the automatic queue depth determination. sys/cam/scsi/scsi_pass.h: Define two new ioctls, CAMIOQUEUE and CAMIOGET, that queue and fetch asynchronous CAM CCBs respectively. Although these ioctls do not have a declared argument, they both take a union ccb pointer. If we declare a size here, the ioctl code in sys/kern/sys_generic.c will malloc and free a buffer for either the CCB or the CCB pointer (depending on how it is declared). Since we have to keep a copy of the CCB (which is fairly large) anyway, having the ioctl malloc and free a CCB for each call is wasteful. sys/cam/scsi/scsi_pass.c: Add asynchronous CCB support. Add two new ioctls, CAMIOQUEUE and CAMIOGET. CAMIOQUEUE adds a CCB to the incoming queue. The CCB is executed immediately (and moved to the active queue) if it is an immediate CCB, but otherwise it will be executed in passstart() when a CCB is available from the transport layer. When CCBs are completed (because they are immediate or passdone() if they are queued), they are put on the done queue. If we get the final close on the device before all pending I/O is complete, all active I/O is moved to the abandoned queue and we increment the peripheral reference count so that the peripheral driver instance doesn't go away before all pending I/O is done. The new passcreatezone() function is called on the first call to the CAMIOQUEUE ioctl on a given device to allocate the UMA zones for I/O requests and S/G list buffers. This may be good to move off to a taskqueue at some point. The new passmemsetup() function allocates memory and scatter/gather lists to hold the user's data, and copies in any data that needs to be written. For virtual pointers (CAM_DATA_VADDR), the kernel buffer is malloced from the new pass(4) driver malloc bucket. For virtual scatter/gather lists (CAM_DATA_SG), buffers are allocated from a new per-pass(9) UMA zone in MAXPHYS-sized chunks. Physical pointers are passed in unchanged. We have support for up to 16 scatter/gather segments (for the user and kernel S/G lists) in the default struct pass_io_req, so requests with longer S/G lists require an extra kernel malloc. The new passcopysglist() function copies a user scatter/gather list to a kernel scatter/gather list. The number of elements in each list may be different, but (obviously) the amount of data stored has to be identical. The new passmemdone() function copies data out for the CAM_DATA_VADDR and CAM_DATA_SG cases. The new passiocleanup() function restores data pointers in user CCBs and frees memory. Add new functions to support kqueue(2)/kevent(2): passreadfilt() tells kevent whether or not the done queue is empty. passkqfilter() adds a knote to our list. passreadfiltdetach() removes a knote from our list. Add a new function, passpoll(), for poll(2)/select(2) to use. Add devstat(9) support for the queued CCB path. sys/cam/ata/ata_da.c: Add support for the BIO_VLIST bio type. sys/cam/cam_ccb.h: Add a new enumeration for the xflags field in the CCB header. (This doesn't change the CCB header, just adds an enumeration to use.) sys/cam/cam_xpt.c: Add a new function, xpt_setup_ccb_flags(), that allows specifying CCB flags. sys/cam/cam_xpt.h: Add a prototype for xpt_setup_ccb_flags(). sys/cam/scsi/scsi_da.c: Add support for BIO_VLIST. sys/dev/md/md.c: Add BIO_VLIST support to md(4). sys/geom/geom_disk.c: Add BIO_VLIST support to the GEOM disk class. Re-factor the I/O size limiting code in g_disk_start() a bit. sys/kern/subr_bus_dma.c: Change _bus_dmamap_load_vlist() to take a starting offset and length. Add a new function, _bus_dmamap_load_pages(), that will load a list of physical pages starting at an offset. Update _bus_dmamap_load_bio() to allow loading BIO_VLIST bios. Allow unmapped I/O to start at an offset. sys/kern/subr_uio.c: Add two new functions, physcopyin_vlist() and physcopyout_vlist(). sys/pc98/include/bus.h: Guard kernel-only parts of the pc98 machine/bus.h header with #ifdef _KERNEL. This allows userland programs to include <machine/bus.h> to get the definition of bus_addr_t and bus_size_t. sys/sys/bio.h: Add a new bio flag, BIO_VLIST. sys/sys/uio.h: Add prototypes for physcopyin_vlist() and physcopyout_vlist(). share/man/man4/pass.4: Document the CAMIOQUEUE and CAMIOGET ioctls. usr.sbin/Makefile: Add camdd. usr.sbin/camdd/Makefile: Add a makefile for camdd(8). usr.sbin/camdd/camdd.8: Man page for camdd(8). usr.sbin/camdd/camdd.c: The new camdd(8) utility. Sponsored by: Spectra Logic MFC after: 1 week
2015-12-03 20:54:55 +00:00
rw_op,
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
/*byte2*/0,
softc->minimum_cmd_size,
/*lba*/bp->bio_pblkno,
/*block_count*/bp->bio_bcount /
softc->params.secsize,
Add asynchronous command support to the pass(4) driver, and the new camdd(8) utility. CCBs may be queued to the driver via the new CAMIOQUEUE ioctl, and completed CCBs may be retrieved via the CAMIOGET ioctl. User processes can use poll(2) or kevent(2) to get notification when I/O has completed. While the existing CAMIOCOMMAND blocking ioctl interface only supports user virtual data pointers in a CCB (generally only one per CCB), the new CAMIOQUEUE ioctl supports user virtual and physical address pointers, as well as user virtual and physical scatter/gather lists. This allows user applications to have more flexibility in their data handling operations. Kernel memory for data transferred via the queued interface is allocated from the zone allocator in MAXPHYS sized chunks, and user data is copied in and out. This is likely faster than the vmapbuf()/vunmapbuf() method used by the CAMIOCOMMAND ioctl in configurations with many processors (there are more TLB shootdowns caused by the mapping/unmapping operation) but may not be as fast as running with unmapped I/O. The new memory handling model for user requests also allows applications to send CCBs with request sizes that are larger than MAXPHYS. The pass(4) driver now limits queued requests to the I/O size listed by the SIM driver in the maxio field in the Path Inquiry (XPT_PATH_INQ) CCB. There are some things things would be good to add: 1. Come up with a way to do unmapped I/O on multiple buffers. Currently the unmapped I/O interface operates on a struct bio, which includes only one address and length. It would be nice to be able to send an unmapped scatter/gather list down to busdma. This would allow eliminating the copy we currently do for data. 2. Add an ioctl to list currently outstanding CCBs in the various queues. 3. Add an ioctl to cancel a request, or use the XPT_ABORT CCB to do that. 4. Test physical address support. Virtual pointers and scatter gather lists have been tested, but I have not yet tested physical addresses or scatter/gather lists. 5. Investigate multiple queue support. At the moment there is one queue of commands per pass(4) device. If multiple processes open the device, they will submit I/O into the same queue and get events for the same completions. This is probably the right model for most applications, but it is something that could be changed later on. Also, add a new utility, camdd(8) that uses the asynchronous pass(4) driver interface. This utility is intended to be a basic data transfer/copy utility, a simple benchmark utility, and an example of how to use the asynchronous pass(4) interface. It can copy data to and from pass(4) devices using any target queue depth, starting offset and blocksize for the input and ouptut devices. It currently only supports SCSI devices, but could be easily extended to support ATA devices. It can also copy data to and from regular files, block devices, tape devices, pipes, stdin, and stdout. It does not support queueing multiple commands to any of those targets, since it uses the standard read(2)/write(2)/writev(2)/readv(2) system calls. The I/O is done by two threads, one for the reader and one for the writer. The reader thread sends completed read requests to the writer thread in strictly sequential order, even if they complete out of order. That could be modified later on for random I/O patterns or slightly out of order I/O. camdd(8) uses kqueue(2)/kevent(2) to get I/O completion events from the pass(4) driver and also to send request notifications internally. For pass(4) devcies, camdd(8) uses a single buffer (CAM_DATA_VADDR) per CAM CCB on the reading side, and a scatter/gather list (CAM_DATA_SG) on the writing side. In addition to testing both interfaces, this makes any potential reblocking of I/O easier. No data is copied between the reader and the writer, but rather the reader's buffers are split into multiple I/O requests or combined into a single I/O request depending on the input and output blocksize. For the file I/O path, camdd(8) also uses a single buffer (read(2), write(2), pread(2) or pwrite(2)) on reads, and a scatter/gather list (readv(2), writev(2), preadv(2), pwritev(2)) on writes. Things that would be nice to do for camdd(8) eventually: 1. Add support for I/O pattern generation. Patterns like all zeros, all ones, LBA-based patterns, random patterns, etc. Right Now you can always use /dev/zero, /dev/random, etc. 2. Add support for a "sink" mode, so we do only reads with no writes. Right now, you can use /dev/null. 3. Add support for automatic queue depth probing, so that we can figure out the right queue depth on the input and output side for maximum throughput. At the moment it defaults to 6. 4. Add support for SATA device passthrough I/O. 5. Add support for random LBAs and/or lengths on the input and output sides. 6. Track average per-I/O latency and busy time. The busy time and latency could also feed in to the automatic queue depth determination. sys/cam/scsi/scsi_pass.h: Define two new ioctls, CAMIOQUEUE and CAMIOGET, that queue and fetch asynchronous CAM CCBs respectively. Although these ioctls do not have a declared argument, they both take a union ccb pointer. If we declare a size here, the ioctl code in sys/kern/sys_generic.c will malloc and free a buffer for either the CCB or the CCB pointer (depending on how it is declared). Since we have to keep a copy of the CCB (which is fairly large) anyway, having the ioctl malloc and free a CCB for each call is wasteful. sys/cam/scsi/scsi_pass.c: Add asynchronous CCB support. Add two new ioctls, CAMIOQUEUE and CAMIOGET. CAMIOQUEUE adds a CCB to the incoming queue. The CCB is executed immediately (and moved to the active queue) if it is an immediate CCB, but otherwise it will be executed in passstart() when a CCB is available from the transport layer. When CCBs are completed (because they are immediate or passdone() if they are queued), they are put on the done queue. If we get the final close on the device before all pending I/O is complete, all active I/O is moved to the abandoned queue and we increment the peripheral reference count so that the peripheral driver instance doesn't go away before all pending I/O is done. The new passcreatezone() function is called on the first call to the CAMIOQUEUE ioctl on a given device to allocate the UMA zones for I/O requests and S/G list buffers. This may be good to move off to a taskqueue at some point. The new passmemsetup() function allocates memory and scatter/gather lists to hold the user's data, and copies in any data that needs to be written. For virtual pointers (CAM_DATA_VADDR), the kernel buffer is malloced from the new pass(4) driver malloc bucket. For virtual scatter/gather lists (CAM_DATA_SG), buffers are allocated from a new per-pass(9) UMA zone in MAXPHYS-sized chunks. Physical pointers are passed in unchanged. We have support for up to 16 scatter/gather segments (for the user and kernel S/G lists) in the default struct pass_io_req, so requests with longer S/G lists require an extra kernel malloc. The new passcopysglist() function copies a user scatter/gather list to a kernel scatter/gather list. The number of elements in each list may be different, but (obviously) the amount of data stored has to be identical. The new passmemdone() function copies data out for the CAM_DATA_VADDR and CAM_DATA_SG cases. The new passiocleanup() function restores data pointers in user CCBs and frees memory. Add new functions to support kqueue(2)/kevent(2): passreadfilt() tells kevent whether or not the done queue is empty. passkqfilter() adds a knote to our list. passreadfiltdetach() removes a knote from our list. Add a new function, passpoll(), for poll(2)/select(2) to use. Add devstat(9) support for the queued CCB path. sys/cam/ata/ata_da.c: Add support for the BIO_VLIST bio type. sys/cam/cam_ccb.h: Add a new enumeration for the xflags field in the CCB header. (This doesn't change the CCB header, just adds an enumeration to use.) sys/cam/cam_xpt.c: Add a new function, xpt_setup_ccb_flags(), that allows specifying CCB flags. sys/cam/cam_xpt.h: Add a prototype for xpt_setup_ccb_flags(). sys/cam/scsi/scsi_da.c: Add support for BIO_VLIST. sys/dev/md/md.c: Add BIO_VLIST support to md(4). sys/geom/geom_disk.c: Add BIO_VLIST support to the GEOM disk class. Re-factor the I/O size limiting code in g_disk_start() a bit. sys/kern/subr_bus_dma.c: Change _bus_dmamap_load_vlist() to take a starting offset and length. Add a new function, _bus_dmamap_load_pages(), that will load a list of physical pages starting at an offset. Update _bus_dmamap_load_bio() to allow loading BIO_VLIST bios. Allow unmapped I/O to start at an offset. sys/kern/subr_uio.c: Add two new functions, physcopyin_vlist() and physcopyout_vlist(). sys/pc98/include/bus.h: Guard kernel-only parts of the pc98 machine/bus.h header with #ifdef _KERNEL. This allows userland programs to include <machine/bus.h> to get the definition of bus_addr_t and bus_size_t. sys/sys/bio.h: Add a new bio flag, BIO_VLIST. sys/sys/uio.h: Add prototypes for physcopyin_vlist() and physcopyout_vlist(). share/man/man4/pass.4: Document the CAMIOQUEUE and CAMIOGET ioctls. usr.sbin/Makefile: Add camdd. usr.sbin/camdd/Makefile: Add a makefile for camdd(8). usr.sbin/camdd/camdd.8: Man page for camdd(8). usr.sbin/camdd/camdd.c: The new camdd(8) utility. Sponsored by: Spectra Logic MFC after: 1 week
2015-12-03 20:54:55 +00:00
data_ptr,
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
/*dxfer_len*/ bp->bio_bcount,
/*sense_len*/SSD_FULL_SIZE,
da_default_timeout * 1000);
break;
Add asynchronous command support to the pass(4) driver, and the new camdd(8) utility. CCBs may be queued to the driver via the new CAMIOQUEUE ioctl, and completed CCBs may be retrieved via the CAMIOGET ioctl. User processes can use poll(2) or kevent(2) to get notification when I/O has completed. While the existing CAMIOCOMMAND blocking ioctl interface only supports user virtual data pointers in a CCB (generally only one per CCB), the new CAMIOQUEUE ioctl supports user virtual and physical address pointers, as well as user virtual and physical scatter/gather lists. This allows user applications to have more flexibility in their data handling operations. Kernel memory for data transferred via the queued interface is allocated from the zone allocator in MAXPHYS sized chunks, and user data is copied in and out. This is likely faster than the vmapbuf()/vunmapbuf() method used by the CAMIOCOMMAND ioctl in configurations with many processors (there are more TLB shootdowns caused by the mapping/unmapping operation) but may not be as fast as running with unmapped I/O. The new memory handling model for user requests also allows applications to send CCBs with request sizes that are larger than MAXPHYS. The pass(4) driver now limits queued requests to the I/O size listed by the SIM driver in the maxio field in the Path Inquiry (XPT_PATH_INQ) CCB. There are some things things would be good to add: 1. Come up with a way to do unmapped I/O on multiple buffers. Currently the unmapped I/O interface operates on a struct bio, which includes only one address and length. It would be nice to be able to send an unmapped scatter/gather list down to busdma. This would allow eliminating the copy we currently do for data. 2. Add an ioctl to list currently outstanding CCBs in the various queues. 3. Add an ioctl to cancel a request, or use the XPT_ABORT CCB to do that. 4. Test physical address support. Virtual pointers and scatter gather lists have been tested, but I have not yet tested physical addresses or scatter/gather lists. 5. Investigate multiple queue support. At the moment there is one queue of commands per pass(4) device. If multiple processes open the device, they will submit I/O into the same queue and get events for the same completions. This is probably the right model for most applications, but it is something that could be changed later on. Also, add a new utility, camdd(8) that uses the asynchronous pass(4) driver interface. This utility is intended to be a basic data transfer/copy utility, a simple benchmark utility, and an example of how to use the asynchronous pass(4) interface. It can copy data to and from pass(4) devices using any target queue depth, starting offset and blocksize for the input and ouptut devices. It currently only supports SCSI devices, but could be easily extended to support ATA devices. It can also copy data to and from regular files, block devices, tape devices, pipes, stdin, and stdout. It does not support queueing multiple commands to any of those targets, since it uses the standard read(2)/write(2)/writev(2)/readv(2) system calls. The I/O is done by two threads, one for the reader and one for the writer. The reader thread sends completed read requests to the writer thread in strictly sequential order, even if they complete out of order. That could be modified later on for random I/O patterns or slightly out of order I/O. camdd(8) uses kqueue(2)/kevent(2) to get I/O completion events from the pass(4) driver and also to send request notifications internally. For pass(4) devcies, camdd(8) uses a single buffer (CAM_DATA_VADDR) per CAM CCB on the reading side, and a scatter/gather list (CAM_DATA_SG) on the writing side. In addition to testing both interfaces, this makes any potential reblocking of I/O easier. No data is copied between the reader and the writer, but rather the reader's buffers are split into multiple I/O requests or combined into a single I/O request depending on the input and output blocksize. For the file I/O path, camdd(8) also uses a single buffer (read(2), write(2), pread(2) or pwrite(2)) on reads, and a scatter/gather list (readv(2), writev(2), preadv(2), pwritev(2)) on writes. Things that would be nice to do for camdd(8) eventually: 1. Add support for I/O pattern generation. Patterns like all zeros, all ones, LBA-based patterns, random patterns, etc. Right Now you can always use /dev/zero, /dev/random, etc. 2. Add support for a "sink" mode, so we do only reads with no writes. Right now, you can use /dev/null. 3. Add support for automatic queue depth probing, so that we can figure out the right queue depth on the input and output side for maximum throughput. At the moment it defaults to 6. 4. Add support for SATA device passthrough I/O. 5. Add support for random LBAs and/or lengths on the input and output sides. 6. Track average per-I/O latency and busy time. The busy time and latency could also feed in to the automatic queue depth determination. sys/cam/scsi/scsi_pass.h: Define two new ioctls, CAMIOQUEUE and CAMIOGET, that queue and fetch asynchronous CAM CCBs respectively. Although these ioctls do not have a declared argument, they both take a union ccb pointer. If we declare a size here, the ioctl code in sys/kern/sys_generic.c will malloc and free a buffer for either the CCB or the CCB pointer (depending on how it is declared). Since we have to keep a copy of the CCB (which is fairly large) anyway, having the ioctl malloc and free a CCB for each call is wasteful. sys/cam/scsi/scsi_pass.c: Add asynchronous CCB support. Add two new ioctls, CAMIOQUEUE and CAMIOGET. CAMIOQUEUE adds a CCB to the incoming queue. The CCB is executed immediately (and moved to the active queue) if it is an immediate CCB, but otherwise it will be executed in passstart() when a CCB is available from the transport layer. When CCBs are completed (because they are immediate or passdone() if they are queued), they are put on the done queue. If we get the final close on the device before all pending I/O is complete, all active I/O is moved to the abandoned queue and we increment the peripheral reference count so that the peripheral driver instance doesn't go away before all pending I/O is done. The new passcreatezone() function is called on the first call to the CAMIOQUEUE ioctl on a given device to allocate the UMA zones for I/O requests and S/G list buffers. This may be good to move off to a taskqueue at some point. The new passmemsetup() function allocates memory and scatter/gather lists to hold the user's data, and copies in any data that needs to be written. For virtual pointers (CAM_DATA_VADDR), the kernel buffer is malloced from the new pass(4) driver malloc bucket. For virtual scatter/gather lists (CAM_DATA_SG), buffers are allocated from a new per-pass(9) UMA zone in MAXPHYS-sized chunks. Physical pointers are passed in unchanged. We have support for up to 16 scatter/gather segments (for the user and kernel S/G lists) in the default struct pass_io_req, so requests with longer S/G lists require an extra kernel malloc. The new passcopysglist() function copies a user scatter/gather list to a kernel scatter/gather list. The number of elements in each list may be different, but (obviously) the amount of data stored has to be identical. The new passmemdone() function copies data out for the CAM_DATA_VADDR and CAM_DATA_SG cases. The new passiocleanup() function restores data pointers in user CCBs and frees memory. Add new functions to support kqueue(2)/kevent(2): passreadfilt() tells kevent whether or not the done queue is empty. passkqfilter() adds a knote to our list. passreadfiltdetach() removes a knote from our list. Add a new function, passpoll(), for poll(2)/select(2) to use. Add devstat(9) support for the queued CCB path. sys/cam/ata/ata_da.c: Add support for the BIO_VLIST bio type. sys/cam/cam_ccb.h: Add a new enumeration for the xflags field in the CCB header. (This doesn't change the CCB header, just adds an enumeration to use.) sys/cam/cam_xpt.c: Add a new function, xpt_setup_ccb_flags(), that allows specifying CCB flags. sys/cam/cam_xpt.h: Add a prototype for xpt_setup_ccb_flags(). sys/cam/scsi/scsi_da.c: Add support for BIO_VLIST. sys/dev/md/md.c: Add BIO_VLIST support to md(4). sys/geom/geom_disk.c: Add BIO_VLIST support to the GEOM disk class. Re-factor the I/O size limiting code in g_disk_start() a bit. sys/kern/subr_bus_dma.c: Change _bus_dmamap_load_vlist() to take a starting offset and length. Add a new function, _bus_dmamap_load_pages(), that will load a list of physical pages starting at an offset. Update _bus_dmamap_load_bio() to allow loading BIO_VLIST bios. Allow unmapped I/O to start at an offset. sys/kern/subr_uio.c: Add two new functions, physcopyin_vlist() and physcopyout_vlist(). sys/pc98/include/bus.h: Guard kernel-only parts of the pc98 machine/bus.h header with #ifdef _KERNEL. This allows userland programs to include <machine/bus.h> to get the definition of bus_addr_t and bus_size_t. sys/sys/bio.h: Add a new bio flag, BIO_VLIST. sys/sys/uio.h: Add prototypes for physcopyin_vlist() and physcopyout_vlist(). share/man/man4/pass.4: Document the CAMIOQUEUE and CAMIOGET ioctls. usr.sbin/Makefile: Add camdd. usr.sbin/camdd/Makefile: Add a makefile for camdd(8). usr.sbin/camdd/camdd.8: Man page for camdd(8). usr.sbin/camdd/camdd.c: The new camdd(8) utility. Sponsored by: Spectra Logic MFC after: 1 week
2015-12-03 20:54:55 +00:00
}
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
case BIO_FLUSH:
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
/*
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
* BIO_FLUSH doesn't currently communicate
* range data, so we synchronize the cache
* over the whole disk. We also force
* ordered tag semantics the flush applies
* to all previously queued I/O.
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
*/
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
scsi_synchronize_cache(&start_ccb->csio,
/*retries*/1,
/*cbfcnp*/dadone,
MSG_ORDERED_Q_TAG,
/*begin_lba*/0,
/*lb_count*/0,
SSD_FULL_SIZE,
da_default_timeout*1000);
break;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
start_ccb->ccb_h.ccb_state = DA_CCB_BUFFER_IO;
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
start_ccb->ccb_h.flags |= CAM_UNLOCKED;
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
out:
LIST_INSERT_HEAD(&softc->pending_ccbs,
&start_ccb->ccb_h, periph_links.le);
/* We expect a unit attention from this device */
if ((softc->flags & DA_FLAG_RETRY_UA) != 0) {
start_ccb->ccb_h.ccb_state |= DA_CCB_RETRY_UA;
softc->flags &= ~DA_FLAG_RETRY_UA;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
start_ccb->ccb_h.ccb_bp = bp;
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
softc->refcount++;
cam_periph_unlock(periph);
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
xpt_action(start_ccb);
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
cam_periph_lock(periph);
softc->refcount--;
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
/* May have more work to do, so ensure we stay scheduled */
daschedule(periph);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
break;
}
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
case DA_STATE_PROBE_RC:
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
{
struct scsi_read_capacity_data *rcap;
Zero scsi_readcapacity allocations so we can really tell if there has been data returned (for some broken residual checking points). MFC after: 3 days
2007-05-18 21:58:07 +00:00
rcap = (struct scsi_read_capacity_data *)
malloc(sizeof(*rcap), M_SCSIDA, M_NOWAIT|M_ZERO);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
if (rcap == NULL) {
printf("dastart: Couldn't malloc read_capacity data\n");
/* da_free_periph??? */
break;
}
Removes essentially unused variables from scsi_da probe setups PR: kern/169835 Reviewed by: pjd (mentor) Approved by: mav MFC after: 2 weeks
2013-01-10 11:28:12 +00:00
scsi_read_capacity(&start_ccb->csio,
Changed scsi_da device requests to use the sysctl tunable value for retry_count and da_default_timeout where their current hardcoded values matched the current default value for said tunables. PR: kern/169976 Reviewed by: pjd (mentor) Approved by: mav
2013-01-10 12:25:00 +00:00
/*retries*/da_retry_count,
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
dadone,
MSG_SIMPLE_Q_TAG,
rcap,
SSD_FULL_SIZE,
/*timeout*/5000);
start_ccb->ccb_h.ccb_bp = NULL;
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
start_ccb->ccb_h.ccb_state = DA_CCB_PROBE_RC;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
xpt_action(start_ccb);
break;
}
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
case DA_STATE_PROBE_RC16:
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
{
struct scsi_read_capacity_data_long *rcaplong;
rcaplong = (struct scsi_read_capacity_data_long *)
Zero scsi_readcapacity allocations so we can really tell if there has been data returned (for some broken residual checking points). MFC after: 3 days
2007-05-18 21:58:07 +00:00
malloc(sizeof(*rcaplong), M_SCSIDA, M_NOWAIT|M_ZERO);
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
if (rcaplong == NULL) {
printf("dastart: Couldn't malloc read_capacity data\n");
/* da_free_periph??? */
break;
}
Removes essentially unused variables from scsi_da probe setups PR: kern/169835 Reviewed by: pjd (mentor) Approved by: mav MFC after: 2 weeks
2013-01-10 11:28:12 +00:00
scsi_read_capacity_16(&start_ccb->csio,
Changed scsi_da device requests to use the sysctl tunable value for retry_count and da_default_timeout where their current hardcoded values matched the current default value for said tunables. PR: kern/169976 Reviewed by: pjd (mentor) Approved by: mav
2013-01-10 12:25:00 +00:00
/*retries*/ da_retry_count,
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
/*cbfcnp*/ dadone,
/*tag_action*/ MSG_SIMPLE_Q_TAG,
/*lba*/ 0,
/*reladr*/ 0,
/*pmi*/ 0,
Add CAM infrastructure to allow reporting when a drive's long read capacity data changes. cam_ccb.h: Add a new advanced information type, CDAI_TYPE_RCAPLONG, for long read capacity data. cam_xpt_internal.h: Add a read capacity data pointer and length to struct cam_ed. cam_xpt.c: Free the read capacity buffer when a device goes away. While we're here, make sure we don't leak memory for other malloced fields in struct cam_ed. scsi_all.c: Update the scsi_read_capacity_16() to take a uint8_t * and a length instead of just a pointer to the parameter data structure. This will hopefully make this function somewhat immune to future changes in the parameter data. scsi_all.h: Add some extra bit definitions to struct scsi_read_capacity_data_long, and bump up the structure size to the full size specified by SBC-3. Change the prototype for scsi_read_capacity_16(). scsi_da.c: Register changes in read capacity data with the transport layer. This allows the transport layer to send out an async notification to interested parties. Update the dasetgeom() API. Use scsi_extract_sense_len() instead of scsi_extract_sense(). scsi_xpt.c: Add support for the new CDAI_TYPE_RCAPLONG advanced information type. Make sure we set the physpath pointer to NULL after freeing it. This allows blindly freeing it in the struct cam_ed destructor. sys/param.h: Bump __FreeBSD_version from 1000005 to 1000006 to make it easier for third party drivers to determine that the read capacity data async notification is available. camcontrol.c, mptutil/mpt_cam.c: Update these for the new scsi_read_capacity_16() argument structure. Sponsored by: Spectra Logic
2012-01-26 18:09:28 +00:00
/*rcap_buf*/ (uint8_t *)rcaplong,
/*rcap_buf_len*/ sizeof(*rcaplong),
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
/*sense_len*/ SSD_FULL_SIZE,
Changed scsi_da device requests to use the sysctl tunable value for retry_count and da_default_timeout where their current hardcoded values matched the current default value for said tunables. PR: kern/169976 Reviewed by: pjd (mentor) Approved by: mav
2013-01-10 12:25:00 +00:00
/*timeout*/ da_default_timeout * 1000);
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
start_ccb->ccb_h.ccb_bp = NULL;
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
start_ccb->ccb_h.ccb_state = DA_CCB_PROBE_RC16;
xpt_action(start_ccb);
break;
}
case DA_STATE_PROBE_LBP:
{
struct scsi_vpd_logical_block_prov *lbp;
if (!scsi_vpd_supported_page(periph, SVPD_LBP)) {
/*
* If we get here we don't support any SBC-3 delete
* methods with UNMAP as the Logical Block Provisioning
* VPD page support is required for devices which
* support it according to T10/1799-D Revision 31
* however older revisions of the spec don't mandate
* this so we currently don't remove these methods
* from the available set.
*/
softc->state = DA_STATE_PROBE_BLK_LIMITS;
goto skipstate;
}
lbp = (struct scsi_vpd_logical_block_prov *)
malloc(sizeof(*lbp), M_SCSIDA, M_NOWAIT|M_ZERO);
if (lbp == NULL) {
printf("dastart: Couldn't malloc lbp data\n");
/* da_free_periph??? */
break;
}
scsi_inquiry(&start_ccb->csio,
/*retries*/da_retry_count,
/*cbfcnp*/dadone,
/*tag_action*/MSG_SIMPLE_Q_TAG,
/*inq_buf*/(u_int8_t *)lbp,
/*inq_len*/sizeof(*lbp),
/*evpd*/TRUE,
/*page_code*/SVPD_LBP,
/*sense_len*/SSD_MIN_SIZE,
/*timeout*/da_default_timeout * 1000);
start_ccb->ccb_h.ccb_bp = NULL;
start_ccb->ccb_h.ccb_state = DA_CCB_PROBE_LBP;
xpt_action(start_ccb);
break;
}
case DA_STATE_PROBE_BLK_LIMITS:
{
struct scsi_vpd_block_limits *block_limits;
if (!scsi_vpd_supported_page(periph, SVPD_BLOCK_LIMITS)) {
/* Not supported skip to next probe */
Update probe flow so that devices with lbp can also disable disksort. Ensure that delete_available is reset so re-probes after a media change, to one with different delete characteristics, will result in the correct methods being flagged as available. Make all ccb state changes use a consistent flow: * free() * xpt_release_ccb() * softc->state = <new state> * xpt_schedule() Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 14:14:57 +00:00
softc->state = DA_STATE_PROBE_BDC;
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
goto skipstate;
}
block_limits = (struct scsi_vpd_block_limits *)
malloc(sizeof(*block_limits), M_SCSIDA, M_NOWAIT|M_ZERO);
if (block_limits == NULL) {
printf("dastart: Couldn't malloc block_limits data\n");
/* da_free_periph??? */
break;
}
scsi_inquiry(&start_ccb->csio,
/*retries*/da_retry_count,
/*cbfcnp*/dadone,
/*tag_action*/MSG_SIMPLE_Q_TAG,
/*inq_buf*/(u_int8_t *)block_limits,
/*inq_len*/sizeof(*block_limits),
/*evpd*/TRUE,
/*page_code*/SVPD_BLOCK_LIMITS,
/*sense_len*/SSD_MIN_SIZE,
/*timeout*/da_default_timeout * 1000);
start_ccb->ccb_h.ccb_bp = NULL;
start_ccb->ccb_h.ccb_state = DA_CCB_PROBE_BLK_LIMITS;
xpt_action(start_ccb);
break;
}
Added automatic detection of non-rotating media which disables the use of BIO queue sorting, hence optimising performance for devices such as SSD's Reviewed by: scottl Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:31:03 +00:00
case DA_STATE_PROBE_BDC:
{
struct scsi_vpd_block_characteristics *bdc;
if (!scsi_vpd_supported_page(periph, SVPD_BDC)) {
softc->state = DA_STATE_PROBE_ATA;
goto skipstate;
}
bdc = (struct scsi_vpd_block_characteristics *)
malloc(sizeof(*bdc), M_SCSIDA, M_NOWAIT|M_ZERO);
if (bdc == NULL) {
printf("dastart: Couldn't malloc bdc data\n");
/* da_free_periph??? */
break;
}
scsi_inquiry(&start_ccb->csio,
/*retries*/da_retry_count,
/*cbfcnp*/dadone,
/*tag_action*/MSG_SIMPLE_Q_TAG,
/*inq_buf*/(u_int8_t *)bdc,
/*inq_len*/sizeof(*bdc),
/*evpd*/TRUE,
/*page_code*/SVPD_BDC,
/*sense_len*/SSD_MIN_SIZE,
/*timeout*/da_default_timeout * 1000);
start_ccb->ccb_h.ccb_bp = NULL;
start_ccb->ccb_h.ccb_state = DA_CCB_PROBE_BDC;
xpt_action(start_ccb);
break;
}
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
case DA_STATE_PROBE_ATA:
{
struct ata_params *ata_params;
Use the existence of ATA Information VPD to determine if we should attempt to query ATA functionality via ATA Pass-Through (16) as this page is defined as "must" for SATL devices, hence indicating that the device is at least likely to support Pass-Through (16). This eliminates errors produced by CTL when ATA Pass-Through (16) fails. Switch ATA probe daerror call to SF_NO_PRINT to avoid errors printing out for devices which return invalid errors. Output details about supported and choosen delete method when verbose booted. Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 14:37:23 +00:00
if (!scsi_vpd_supported_page(periph, SVPD_ATA_INFORMATION)) {
daprobedone(periph, start_ccb);
break;
}
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
ata_params = (struct ata_params*)
malloc(sizeof(*ata_params), M_SCSIDA, M_NOWAIT|M_ZERO);
if (ata_params == NULL) {
printf("dastart: Couldn't malloc ata_params data\n");
/* da_free_periph??? */
break;
}
scsi_ata_identify(&start_ccb->csio,
/*retries*/da_retry_count,
/*cbfcnp*/dadone,
/*tag_action*/MSG_SIMPLE_Q_TAG,
/*data_ptr*/(u_int8_t *)ata_params,
/*dxfer_len*/sizeof(*ata_params),
/*sense_len*/SSD_FULL_SIZE,
/*timeout*/da_default_timeout * 1000);
start_ccb->ccb_h.ccb_bp = NULL;
start_ccb->ccb_h.ccb_state = DA_CCB_PROBE_ATA;
xpt_action(start_ccb);
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
break;
}
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
}
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
/*
* In each of the methods below, while its the caller's
* responsibility to ensure the request will fit into a
* single device request, we might have changed the delete
* method due to the device incorrectly advertising either
* its supported methods or limits.
*
* To prevent this causing further issues we validate the
* against the methods limits, and warn which would
* otherwise be unnecessary.
*/
static void
da_delete_unmap(struct cam_periph *periph, union ccb *ccb, struct bio *bp)
{
struct da_softc *softc = (struct da_softc *)periph->softc;;
struct bio *bp1;
uint8_t *buf = softc->unmap_buf;
uint64_t lba, lastlba = (uint64_t)-1;
Mostly revert r260267 and hopefully really fix the original problem. The latest draft of SBC-3 tells: "A MAXIMUM UNMAP LBA COUNT field set to a non-zero value indicates the maximum number of LBAs that may be unmapped by an UNMAP command." To me it does not sound like that limit is set per single descriptor, but rather per all command. And I have at least one device that behaves exactly that way. This patch fixes the problem there. MFC after: 1 week
2014-01-22 22:19:53 +00:00
uint64_t totalcount = 0;
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
uint64_t count;
uint32_t lastcount = 0, c;
uint32_t off, ranges = 0;
/*
* Currently this doesn't take the UNMAP
* Granularity and Granularity Alignment
* fields into account.
*
* This could result in both unoptimal unmap
* requests as as well as UNMAP calls unmapping
* fewer LBA's than requested.
*/
softc->delete_running = 1;
bzero(softc->unmap_buf, sizeof(softc->unmap_buf));
bp1 = bp;
do {
bioq_remove(&softc->delete_queue, bp1);
if (bp1 != bp)
bioq_insert_tail(&softc->delete_run_queue, bp1);
lba = bp1->bio_pblkno;
count = bp1->bio_bcount / softc->params.secsize;
/* Try to extend the previous range. */
if (lba == lastlba) {
Mostly revert r260267 and hopefully really fix the original problem. The latest draft of SBC-3 tells: "A MAXIMUM UNMAP LBA COUNT field set to a non-zero value indicates the maximum number of LBAs that may be unmapped by an UNMAP command." To me it does not sound like that limit is set per single descriptor, but rather per all command. And I have at least one device that behaves exactly that way. This patch fixes the problem there. MFC after: 1 week
2014-01-22 22:19:53 +00:00
c = omin(count, UNMAP_RANGE_MAX - lastcount);
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
lastcount += c;
off = ((ranges - 1) * UNMAP_RANGE_SIZE) +
UNMAP_HEAD_SIZE;
scsi_ulto4b(lastcount, &buf[off + 8]);
count -= c;
lba +=c;
Mostly revert r260267 and hopefully really fix the original problem. The latest draft of SBC-3 tells: "A MAXIMUM UNMAP LBA COUNT field set to a non-zero value indicates the maximum number of LBAs that may be unmapped by an UNMAP command." To me it does not sound like that limit is set per single descriptor, but rather per all command. And I have at least one device that behaves exactly that way. This patch fixes the problem there. MFC after: 1 week
2014-01-22 22:19:53 +00:00
totalcount += c;
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
}
while (count > 0) {
Mostly revert r260267 and hopefully really fix the original problem. The latest draft of SBC-3 tells: "A MAXIMUM UNMAP LBA COUNT field set to a non-zero value indicates the maximum number of LBAs that may be unmapped by an UNMAP command." To me it does not sound like that limit is set per single descriptor, but rather per all command. And I have at least one device that behaves exactly that way. This patch fixes the problem there. MFC after: 1 week
2014-01-22 22:19:53 +00:00
c = omin(count, UNMAP_RANGE_MAX);
if (totalcount + c > softc->unmap_max_lba ||
ranges >= softc->unmap_max_ranges) {
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
xpt_print(periph->path,
Mostly revert r260267 and hopefully really fix the original problem. The latest draft of SBC-3 tells: "A MAXIMUM UNMAP LBA COUNT field set to a non-zero value indicates the maximum number of LBAs that may be unmapped by an UNMAP command." To me it does not sound like that limit is set per single descriptor, but rather per all command. And I have at least one device that behaves exactly that way. This patch fixes the problem there. MFC after: 1 week
2014-01-22 22:19:53 +00:00
"%s issuing short delete %ld > %ld"
"|| %d >= %d",
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
da_delete_method_desc[softc->delete_method],
Mostly revert r260267 and hopefully really fix the original problem. The latest draft of SBC-3 tells: "A MAXIMUM UNMAP LBA COUNT field set to a non-zero value indicates the maximum number of LBAs that may be unmapped by an UNMAP command." To me it does not sound like that limit is set per single descriptor, but rather per all command. And I have at least one device that behaves exactly that way. This patch fixes the problem there. MFC after: 1 week
2014-01-22 22:19:53 +00:00
totalcount + c, softc->unmap_max_lba,
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
ranges, softc->unmap_max_ranges);
break;
}
off = (ranges * UNMAP_RANGE_SIZE) + UNMAP_HEAD_SIZE;
scsi_u64to8b(lba, &buf[off + 0]);
scsi_ulto4b(c, &buf[off + 8]);
lba += c;
Mostly revert r260267 and hopefully really fix the original problem. The latest draft of SBC-3 tells: "A MAXIMUM UNMAP LBA COUNT field set to a non-zero value indicates the maximum number of LBAs that may be unmapped by an UNMAP command." To me it does not sound like that limit is set per single descriptor, but rather per all command. And I have at least one device that behaves exactly that way. This patch fixes the problem there. MFC after: 1 week
2014-01-22 22:19:53 +00:00
totalcount += c;
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
ranges++;
count -= c;
lastcount = c;
}
lastlba = lba;
bp1 = bioq_first(&softc->delete_queue);
if (bp1 == NULL || ranges >= softc->unmap_max_ranges ||
Mostly revert r260267 and hopefully really fix the original problem. The latest draft of SBC-3 tells: "A MAXIMUM UNMAP LBA COUNT field set to a non-zero value indicates the maximum number of LBAs that may be unmapped by an UNMAP command." To me it does not sound like that limit is set per single descriptor, but rather per all command. And I have at least one device that behaves exactly that way. This patch fixes the problem there. MFC after: 1 week
2014-01-22 22:19:53 +00:00
totalcount + bp1->bio_bcount /
softc->params.secsize > softc->unmap_max_lba)
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
break;
} while (1);
scsi_ulto2b(ranges * 16 + 6, &buf[0]);
scsi_ulto2b(ranges * 16, &buf[2]);
scsi_unmap(&ccb->csio,
/*retries*/da_retry_count,
/*cbfcnp*/dadone,
/*tag_action*/MSG_SIMPLE_Q_TAG,
/*byte2*/0,
/*data_ptr*/ buf,
/*dxfer_len*/ ranges * 16 + 8,
/*sense_len*/SSD_FULL_SIZE,
da_default_timeout * 1000);
ccb->ccb_h.ccb_state = DA_CCB_DELETE;
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
ccb->ccb_h.flags |= CAM_UNLOCKED;
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
}
static void
da_delete_trim(struct cam_periph *periph, union ccb *ccb, struct bio *bp)
{
struct da_softc *softc = (struct da_softc *)periph->softc;
struct bio *bp1;
uint8_t *buf = softc->unmap_buf;
uint64_t lastlba = (uint64_t)-1;
uint64_t count;
uint64_t lba;
uint32_t lastcount = 0, c, requestcount;
int ranges = 0, off, block_count;
softc->delete_running = 1;
bzero(softc->unmap_buf, sizeof(softc->unmap_buf));
bp1 = bp;
do {
bioq_remove(&softc->delete_queue, bp1);
if (bp1 != bp)
bioq_insert_tail(&softc->delete_run_queue, bp1);
lba = bp1->bio_pblkno;
count = bp1->bio_bcount / softc->params.secsize;
requestcount = count;
/* Try to extend the previous range. */
if (lba == lastlba) {
Fix several potential overflows in UNMAP code. MFC after: 1 week
2015-01-26 15:47:08 +00:00
c = omin(count, ATA_DSM_RANGE_MAX - lastcount);
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
lastcount += c;
off = (ranges - 1) * 8;
buf[off + 6] = lastcount & 0xff;
buf[off + 7] = (lastcount >> 8) & 0xff;
count -= c;
lba += c;
}
while (count > 0) {
Fix several potential overflows in UNMAP code. MFC after: 1 week
2015-01-26 15:47:08 +00:00
c = omin(count, ATA_DSM_RANGE_MAX);
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
off = ranges * 8;
buf[off + 0] = lba & 0xff;
buf[off + 1] = (lba >> 8) & 0xff;
buf[off + 2] = (lba >> 16) & 0xff;
buf[off + 3] = (lba >> 24) & 0xff;
buf[off + 4] = (lba >> 32) & 0xff;
buf[off + 5] = (lba >> 40) & 0xff;
buf[off + 6] = c & 0xff;
buf[off + 7] = (c >> 8) & 0xff;
lba += c;
ranges++;
count -= c;
lastcount = c;
if (count != 0 && ranges == softc->trim_max_ranges) {
xpt_print(periph->path,
Correct short delete issue in SCSI UNMAP support Correct missing \n's in xpt_print's Correct incorrect count being passed to short delete xpt_print MFC after: 1 week
2014-01-04 17:52:43 +00:00
"%s issuing short delete %ld > %ld\n",
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
da_delete_method_desc[softc->delete_method],
requestcount,
(softc->trim_max_ranges - ranges) *
ATA_DSM_RANGE_MAX);
break;
}
}
lastlba = lba;
bp1 = bioq_first(&softc->delete_queue);
if (bp1 == NULL || bp1->bio_bcount / softc->params.secsize >
(softc->trim_max_ranges - ranges) * ATA_DSM_RANGE_MAX)
break;
} while (1);
block_count = (ranges + ATA_DSM_BLK_RANGES - 1) / ATA_DSM_BLK_RANGES;
scsi_ata_trim(&ccb->csio,
/*retries*/da_retry_count,
/*cbfcnp*/dadone,
/*tag_action*/MSG_SIMPLE_Q_TAG,
block_count,
/*data_ptr*/buf,
/*dxfer_len*/block_count * ATA_DSM_BLK_SIZE,
/*sense_len*/SSD_FULL_SIZE,
da_default_timeout * 1000);
ccb->ccb_h.ccb_state = DA_CCB_DELETE;
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
ccb->ccb_h.flags |= CAM_UNLOCKED;
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
}
/*
* We calculate ws_max_blks here based off d_delmaxsize instead
* of using softc->ws_max_blks as it is absolute max for the
Correct short delete issue in SCSI UNMAP support Correct missing \n's in xpt_print's Correct incorrect count being passed to short delete xpt_print MFC after: 1 week
2014-01-04 17:52:43 +00:00
* device not the protocol max which may well be lower.
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
*/
static void
da_delete_ws(struct cam_periph *periph, union ccb *ccb, struct bio *bp)
{
struct da_softc *softc;
struct bio *bp1;
uint64_t ws_max_blks;
uint64_t lba;
uint64_t count; /* forward compat with WS32 */
softc = (struct da_softc *)periph->softc;
ws_max_blks = softc->disk->d_delmaxsize / softc->params.secsize;
softc->delete_running = 1;
lba = bp->bio_pblkno;
count = 0;
bp1 = bp;
do {
bioq_remove(&softc->delete_queue, bp1);
if (bp1 != bp)
bioq_insert_tail(&softc->delete_run_queue, bp1);
count += bp1->bio_bcount / softc->params.secsize;
if (count > ws_max_blks) {
xpt_print(periph->path,
Correct short delete issue in SCSI UNMAP support Correct missing \n's in xpt_print's Correct incorrect count being passed to short delete xpt_print MFC after: 1 week
2014-01-04 17:52:43 +00:00
"%s issuing short delete %ld > %ld\n",
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
da_delete_method_desc[softc->delete_method],
count, ws_max_blks);
Fix several potential overflows in UNMAP code. MFC after: 1 week
2015-01-26 15:47:08 +00:00
count = omin(count, ws_max_blks);
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
break;
}
bp1 = bioq_first(&softc->delete_queue);
if (bp1 == NULL || lba + count != bp1->bio_pblkno ||
count + bp1->bio_bcount /
softc->params.secsize > ws_max_blks)
break;
} while (1);
scsi_write_same(&ccb->csio,
/*retries*/da_retry_count,
/*cbfcnp*/dadone,
/*tag_action*/MSG_SIMPLE_Q_TAG,
/*byte2*/softc->delete_method ==
DA_DELETE_ZERO ? 0 : SWS_UNMAP,
softc->delete_method == DA_DELETE_WS16 ? 16 : 10,
/*lba*/lba,
/*block_count*/count,
/*data_ptr*/ __DECONST(void *, zero_region),
/*dxfer_len*/ softc->params.secsize,
/*sense_len*/SSD_FULL_SIZE,
da_default_timeout * 1000);
ccb->ccb_h.ccb_state = DA_CCB_DELETE;
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
ccb->ccb_h.flags |= CAM_UNLOCKED;
Refactor the various delete methods out of dastart(). Cleans up a bunch of style and adds more modularity and clarity. Obtained from: Netflix MFC after: 3 days
2013-07-12 00:50:25 +00:00
}
Automatically detect devices that do not support READ(6)/WRITE(6) and upgrade to using 10 byte cdbs. As far as I tested, this works efficiently for most of the SBP-II/Firewire devices but most of the umass devices still need ad-hoc work around because umass-sim doesn't return any SCSI errors. A sysctl nob is also added for the last resort. I hope we don't need DA_Q_NO_6_BYTE quirks anymore. Reviewed by: gibbs MFC after: 1 week
2002-03-23 18:18:02 +00:00
static int
cmd6workaround(union ccb *ccb)
{
struct scsi_rw_6 cmd6;
struct scsi_rw_10 *cmd10;
struct da_softc *softc;
Updates to cmd6workaround: * Only update cdb in place if not CDB_POINTER * Correctly check for QFRZ before restarting CCB * More accurate printf message * style(9) changes at end Patch tested successfully on Maxtor 4 G120J6 GAK8.
2002-08-16 22:05:19 +00:00
u_int8_t *cdb;
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
struct bio *bp;
Updates to cmd6workaround: * Only update cdb in place if not CDB_POINTER * Correctly check for QFRZ before restarting CCB * More accurate printf message * style(9) changes at end Patch tested successfully on Maxtor 4 G120J6 GAK8.
2002-08-16 22:05:19 +00:00
int frozen;
Automatically detect devices that do not support READ(6)/WRITE(6) and upgrade to using 10 byte cdbs. As far as I tested, this works efficiently for most of the SBP-II/Firewire devices but most of the umass devices still need ad-hoc work around because umass-sim doesn't return any SCSI errors. A sysctl nob is also added for the last resort. I hope we don't need DA_Q_NO_6_BYTE quirks anymore. Reviewed by: gibbs MFC after: 1 week
2002-03-23 18:18:02 +00:00
Updates to cmd6workaround: * Only update cdb in place if not CDB_POINTER * Correctly check for QFRZ before restarting CCB * More accurate printf message * style(9) changes at end Patch tested successfully on Maxtor 4 G120J6 GAK8.
2002-08-16 22:05:19 +00:00
cdb = ccb->csio.cdb_io.cdb_bytes;
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
softc = (struct da_softc *)xpt_path_periph(ccb->ccb_h.path)->softc;
if (ccb->ccb_h.ccb_state == DA_CCB_DELETE) {
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
da_delete_methods old_method = softc->delete_method;
/*
* Typically there are two reasons for failure here
* 1. Delete method was detected as supported but isn't
* 2. Delete failed due to invalid params e.g. too big
*
* While we will attempt to choose an alternative delete method
* this may result in short deletes if the existing delete
* requests from geom are big for the new method choosen.
*
* This method assumes that the error which triggered this
* will not retry the io otherwise a panic will occur
*/
dadeleteflag(softc, old_method, 0);
dadeletemethodchoose(softc, DA_DELETE_DISABLE);
if (softc->delete_method == DA_DELETE_DISABLE)
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
xpt_print(ccb->ccb_h.path,
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
"%s failed, disabling BIO_DELETE\n",
da_delete_method_desc[old_method]);
else
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
xpt_print(ccb->ccb_h.path,
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
"%s failed, switching to %s BIO_DELETE\n",
da_delete_method_desc[old_method],
da_delete_method_desc[softc->delete_method]);
Adds the ability to enable / disable sorting of BIO requests queued within CAM. This can significantly improve performance particularly for SSDs which don't suffer from seek latencies. The sysctl / tunable kern.cam.sort_io_queues provides the systems default setting where:- 0 = queued BIOs are NOT sorted 1 = queued BIOs are sorted (default) Each device gets its own sysctl kern.cam.<type>.<id>.sort_io_queue Valid values are:- -1 = use system default (default) 0 = queued BIOs are NOT sorted 1 = queued BIOs are sorted Note: Additional patch will look to add automatic use of none sorted queues for none rotating media e.g. SSD's Reviewed by: scottl Approved by: pjd (mentor) MFC after: 2 weeks
2013-03-29 22:58:15 +00:00
The code that combines adjacent ranges for BIO_DELETEs to optimize trims to the device assumes the list is sorted. Don't apply the optimization of not sorting the queue when we have SSDs to the delete_queue, since it causes more discard traffic to the drive. While one could argue that the higher levels should coalesce the trims, that's not done today, so some optimization at this level is needed. CR: https://phabric.freebsd.org/D142
2014-06-05 17:13:42 +00:00
while ((bp = bioq_takefirst(&softc->delete_run_queue)) != NULL)
bioq_disksort(&softc->delete_queue, bp);
bioq_disksort(&softc->delete_queue,
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
(struct bio *)ccb->ccb_h.ccb_bp);
ccb->ccb_h.ccb_bp = NULL;
return (0);
}
Automatically detect devices that do not support READ(6)/WRITE(6) and upgrade to using 10 byte cdbs. As far as I tested, this works efficiently for most of the SBP-II/Firewire devices but most of the umass devices still need ad-hoc work around because umass-sim doesn't return any SCSI errors. A sysctl nob is also added for the last resort. I hope we don't need DA_Q_NO_6_BYTE quirks anymore. Reviewed by: gibbs MFC after: 1 week
2002-03-23 18:18:02 +00:00
Detect unsupported PREVENT ALLOW MEDIUM REMOVAL and SYNCHRONIZE CACHE(10) to not spam devices with useless commands and logs with errors.
2013-07-27 22:19:34 +00:00
/* Detect unsupported PREVENT ALLOW MEDIUM REMOVAL. */
if ((ccb->ccb_h.flags & CAM_CDB_POINTER) == 0 &&
(*cdb == PREVENT_ALLOW) &&
(softc->quirks & DA_Q_NO_PREVENT) == 0) {
if (bootverbose)
xpt_print(ccb->ccb_h.path,
"PREVENT ALLOW MEDIUM REMOVAL not supported.\n");
softc->quirks |= DA_Q_NO_PREVENT;
return (0);
}
/* Detect unsupported SYNCHRONIZE CACHE(10). */
if ((ccb->ccb_h.flags & CAM_CDB_POINTER) == 0 &&
(*cdb == SYNCHRONIZE_CACHE) &&
(softc->quirks & DA_Q_NO_SYNC_CACHE) == 0) {
if (bootverbose)
xpt_print(ccb->ccb_h.path,
"SYNCHRONIZE CACHE(10) not supported.\n");
softc->quirks |= DA_Q_NO_SYNC_CACHE;
Improve r253721 by reporting detected lack of BIO_FLUSH support to GEOM. That prevents more of such requests from coming and errors from logging.
2013-08-07 08:20:11 +00:00
softc->disk->d_flags &= ~DISKFLAG_CANFLUSHCACHE;
Detect unsupported PREVENT ALLOW MEDIUM REMOVAL and SYNCHRONIZE CACHE(10) to not spam devices with useless commands and logs with errors.
2013-07-27 22:19:34 +00:00
return (0);
}
Updates to cmd6workaround: * Only update cdb in place if not CDB_POINTER * Correctly check for QFRZ before restarting CCB * More accurate printf message * style(9) changes at end Patch tested successfully on Maxtor 4 G120J6 GAK8.
2002-08-16 22:05:19 +00:00
/* Translation only possible if CDB is an array and cmd is R/W6 */
if ((ccb->ccb_h.flags & CAM_CDB_POINTER) != 0 ||
(*cdb != READ_6 && *cdb != WRITE_6))
Automatically detect devices that do not support READ(6)/WRITE(6) and upgrade to using 10 byte cdbs. As far as I tested, this works efficiently for most of the SBP-II/Firewire devices but most of the umass devices still need ad-hoc work around because umass-sim doesn't return any SCSI errors. A sysctl nob is also added for the last resort. I hope we don't need DA_Q_NO_6_BYTE quirks anymore. Reviewed by: gibbs MFC after: 1 week
2002-03-23 18:18:02 +00:00
return 0;
Add an xpt_print function to reduce most of the xpt_print_path/printf pairs. Convert the core code to use it. Reviewed by: scsi (various folks weighing in with good advice) MFC after: 2 weeks
2006-12-05 07:45:28 +00:00
xpt_print(ccb->ccb_h.path, "READ(6)/WRITE(6) not supported, "
"increasing minimum_cmd_size to 10.\n");
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
softc->minimum_cmd_size = 10;
Automatically detect devices that do not support READ(6)/WRITE(6) and upgrade to using 10 byte cdbs. As far as I tested, this works efficiently for most of the SBP-II/Firewire devices but most of the umass devices still need ad-hoc work around because umass-sim doesn't return any SCSI errors. A sysctl nob is also added for the last resort. I hope we don't need DA_Q_NO_6_BYTE quirks anymore. Reviewed by: gibbs MFC after: 1 week
2002-03-23 18:18:02 +00:00
Updates to cmd6workaround: * Only update cdb in place if not CDB_POINTER * Correctly check for QFRZ before restarting CCB * More accurate printf message * style(9) changes at end Patch tested successfully on Maxtor 4 G120J6 GAK8.
2002-08-16 22:05:19 +00:00
bcopy(cdb, &cmd6, sizeof(struct scsi_rw_6));
cmd10 = (struct scsi_rw_10 *)cdb;
Automatically detect devices that do not support READ(6)/WRITE(6) and upgrade to using 10 byte cdbs. As far as I tested, this works efficiently for most of the SBP-II/Firewire devices but most of the umass devices still need ad-hoc work around because umass-sim doesn't return any SCSI errors. A sysctl nob is also added for the last resort. I hope we don't need DA_Q_NO_6_BYTE quirks anymore. Reviewed by: gibbs MFC after: 1 week
2002-03-23 18:18:02 +00:00
cmd10->opcode = (cmd6.opcode == READ_6) ? READ_10 : WRITE_10;
cmd10->byte2 = 0;
scsi_ulto4b(scsi_3btoul(cmd6.addr), cmd10->addr);
cmd10->reserved = 0;
scsi_ulto2b(cmd6.length, cmd10->length);
cmd10->control = cmd6.control;
Updates to cmd6workaround: * Only update cdb in place if not CDB_POINTER * Correctly check for QFRZ before restarting CCB * More accurate printf message * style(9) changes at end Patch tested successfully on Maxtor 4 G120J6 GAK8.
2002-08-16 22:05:19 +00:00
ccb->csio.cdb_len = sizeof(*cmd10);
Automatically detect devices that do not support READ(6)/WRITE(6) and upgrade to using 10 byte cdbs. As far as I tested, this works efficiently for most of the SBP-II/Firewire devices but most of the umass devices still need ad-hoc work around because umass-sim doesn't return any SCSI errors. A sysctl nob is also added for the last resort. I hope we don't need DA_Q_NO_6_BYTE quirks anymore. Reviewed by: gibbs MFC after: 1 week
2002-03-23 18:18:02 +00:00
Updates to cmd6workaround: * Only update cdb in place if not CDB_POINTER * Correctly check for QFRZ before restarting CCB * More accurate printf message * style(9) changes at end Patch tested successfully on Maxtor 4 G120J6 GAK8.
2002-08-16 22:05:19 +00:00
/* Requeue request, unfreezing queue if necessary */
frozen = (ccb->ccb_h.status & CAM_DEV_QFRZN) != 0;
Automatically detect devices that do not support READ(6)/WRITE(6) and upgrade to using 10 byte cdbs. As far as I tested, this works efficiently for most of the SBP-II/Firewire devices but most of the umass devices still need ad-hoc work around because umass-sim doesn't return any SCSI errors. A sysctl nob is also added for the last resort. I hope we don't need DA_Q_NO_6_BYTE quirks anymore. Reviewed by: gibbs MFC after: 1 week
2002-03-23 18:18:02 +00:00
ccb->ccb_h.status = CAM_REQUEUE_REQ;
xpt_action(ccb);
Updates to cmd6workaround: * Only update cdb in place if not CDB_POINTER * Correctly check for QFRZ before restarting CCB * More accurate printf message * style(9) changes at end Patch tested successfully on Maxtor 4 G120J6 GAK8.
2002-08-16 22:05:19 +00:00
if (frozen) {
Automatically detect devices that do not support READ(6)/WRITE(6) and upgrade to using 10 byte cdbs. As far as I tested, this works efficiently for most of the SBP-II/Firewire devices but most of the umass devices still need ad-hoc work around because umass-sim doesn't return any SCSI errors. A sysctl nob is also added for the last resort. I hope we don't need DA_Q_NO_6_BYTE quirks anymore. Reviewed by: gibbs MFC after: 1 week
2002-03-23 18:18:02 +00:00
cam_release_devq(ccb->ccb_h.path,
/*relsim_flags*/0,
/*reduction*/0,
/*timeout*/0,
/*getcount_only*/0);
Updates to cmd6workaround: * Only update cdb in place if not CDB_POINTER * Correctly check for QFRZ before restarting CCB * More accurate printf message * style(9) changes at end Patch tested successfully on Maxtor 4 G120J6 GAK8.
2002-08-16 22:05:19 +00:00
}
Automatically detect devices that do not support READ(6)/WRITE(6) and upgrade to using 10 byte cdbs. As far as I tested, this works efficiently for most of the SBP-II/Firewire devices but most of the umass devices still need ad-hoc work around because umass-sim doesn't return any SCSI errors. A sysctl nob is also added for the last resort. I hope we don't need DA_Q_NO_6_BYTE quirks anymore. Reviewed by: gibbs MFC after: 1 week
2002-03-23 18:18:02 +00:00
return (ERESTART);
}
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
static void
dadone(struct cam_periph *periph, union ccb *done_ccb)
{
struct da_softc *softc;
struct ccb_scsiio *csio;
MFp4: - Reduce code duplication in ATA XPT and PMP driver. - Move PIO size setting from ada driver to ATA XPT. It is XPT business to negotiate transfer details. ada driver is now stateless. - Report PIO size to SIM. It is required for correct PATA SIM operation. - Tune PMP scan timings. It workarounds some problems with SiI. - If reset hapens during PMP initialization - restart it. - Introduce early-initialized periph drivers, which are used during initial scan process. Use it for xpt, probe, aprobe and pmp. It gives pmp chance to finish scan before mountroot and numerate devices in right order.
2009-10-31 10:43:38 +00:00
u_int32_t priority;
Do not schedule periph for payload/TUR requests if reprobe is in progress to avoid sending extra READ CAPACITY requests by dastart(). Schedule periph again on reprobe completion, or otherwise it may stuck indefinitely long. This should fix USB explore thread hanging on device unplug, waiting for periph destruction. Reported by: hselasky
2013-01-11 16:10:11 +00:00
da_ccb_state state;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
softc = (struct da_softc *)periph->softc;
MFp4: - Reduce code duplication in ATA XPT and PMP driver. - Move PIO size setting from ada driver to ATA XPT. It is XPT business to negotiate transfer details. ada driver is now stateless. - Report PIO size to SIM. It is required for correct PATA SIM operation. - Tune PMP scan timings. It workarounds some problems with SiI. - If reset hapens during PMP initialization - restart it. - Introduce early-initialized periph drivers, which are used during initial scan process. Use it for xpt, probe, aprobe and pmp. It gives pmp chance to finish scan before mountroot and numerate devices in right order.
2009-10-31 10:43:38 +00:00
priority = done_ccb->ccb_h.pinfo.priority;
Tune and add some missing CAM_DEBUG() points for better consistency.
2012-06-05 09:45:42 +00:00
CAM_DEBUG(periph->path, CAM_DEBUG_TRACE, ("dadone\n"));
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
csio = &done_ccb->csio;
Do not schedule periph for payload/TUR requests if reprobe is in progress to avoid sending extra READ CAPACITY requests by dastart(). Schedule periph again on reprobe completion, or otherwise it may stuck indefinitely long. This should fix USB explore thread hanging on device unplug, waiting for periph destruction. Reported by: hselasky
2013-01-11 16:10:11 +00:00
state = csio->ccb_h.ccb_state & DA_CCB_TYPE_MASK;
switch (state) {
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
case DA_CCB_BUFFER_IO:
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
case DA_CCB_DELETE:
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
{
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
struct bio *bp, *bp1;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
cam_periph_lock(periph);
Complete the bio/buf divorce for all code below devfs::strategy Exceptions: Vinum untouched. This means that it cannot be compiled. Greg Lehey is on the case. CCD not converted yet, casts to struct buf (still safe) atapi-cd casts to struct buf to examine B_PHYS
2000-04-15 05:54:02 +00:00
bp = (struct bio *)done_ccb->ccb_h.ccb_bp;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
if ((done_ccb->ccb_h.status & CAM_STATUS_MASK) != CAM_REQ_CMP) {
int error;
int sf;
sys/cam/scsi/scsi_da.c: - Only attempt the closing synchronize cache on a disk if it is still there. - When a device is lost, report the number of outstanding I/Os as they are drained. - When a device is lost, return any unprocessed bios with ENXIO instead of EIO. - Filter asynchronous events, but always allow cam_periph_async() to see them too. Sponsored by: Spectra Logic Corporation
2011-06-14 16:05:00 +00:00
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
if ((csio->ccb_h.ccb_state & DA_CCB_RETRY_UA) != 0)
sf = SF_RETRY_UA;
else
sf = 0;
Rewrite of the CAM error recovery code. Some of the major changes include: - The SCSI error handling portion of cam_periph_error() has been broken out into a number of subfunctions to better modularize the code that handles the hierarchy of SCSI errors. As a result, the code is now much easier to read. - String handling and error printing has been significantly revamped. We now use sbufs to do string formatting instead of using printfs (for the kernel) and snprintf/strncat (for userland) as before. There is a new catchall error printing routine, cam_error_print() and its string-based counterpart, cam_error_string() that allow the kernel and userland applications to pass in a CCB and have errors printed out properly, whether or not they're SCSI errors. Among other things, this helped eliminate a fair amount of duplicate code in camcontrol. We now print out more information than before, including the CAM status and SCSI status and the error recovery action taken to remedy the problem. - sbufs are now available in userland, via libsbuf. This change was necessary since most of the error printing code is shared between libcam and the kernel. - A new transfer settings interface is included in this checkin. This code is #ifdef'ed out, and is primarily intended to aid discussion with HBA driver authors on the final form the interface should take. There is example code in the ahc(4) driver that implements the HBA driver side of the new interface. The new transfer settings code won't be enabled until we're ready to switch all HBA drivers over to the new interface. src/Makefile.inc1, lib/Makefile: Add libsbuf. It must be built before libcam, since libcam uses sbuf routines. libcam/Makefile: libcam now depends on libsbuf. libsbuf/Makefile: Add a makefile for libsbuf. This pulls in the sbuf sources from sys/kern. bsd.libnames.mk: Add LIBSBUF. camcontrol/Makefile: Add -lsbuf. Since camcontrol is statically linked, we can't depend on the dynamic linker to pull in libsbuf. camcontrol.c: Use cam_error_print() instead of checking for CAM_SCSI_STATUS_ERROR on every failed CCB. sbuf.9: Change the prototypes for sbuf_cat() and sbuf_cpy() so that the source string is now a const char *. This is more in line wth the standard system string functions, and helps eliminate warnings when dealing with a const source buffer. Fix a typo. cam.c: Add description strings for the various CAM error status values, as well as routines to look up those strings. Add new cam_error_string() and cam_error_print() routines for userland and the kernel. cam.h: Add a new CAM flag, CAM_RETRY_SELTO. Add enumerated types for the various options available with cam_error_print() and cam_error_string(). cam_ccb.h: Add new transfer negotiation structures/types. Change inq_len in the ccb_getdev structure to be "reserved". This field has never been filled in, and will be removed when we next bump the CAM version. cam_debug.h: Fix typo. cam_periph.c: Modularize cam_periph_error(). The SCSI error handling part of cam_periph_error() is now in camperiphscsistatuserror() and camperiphscsisenseerror(). In cam_periph_lock(), increase the reference count on the periph while we wait for our lock attempt to succeed so that the periph won't go away while we're sleeping. cam_xpt.c: Add new transfer negotiation code. (ifdefed out) Add a new function, xpt_path_string(). This is a string/sbuf analog to xpt_print_path(). scsi_all.c: Revamp string handing and error printing code. We now use sbufs for much of the string formatting code. More of that code is shared between userland the kernel. scsi_all.h: Get rid of SS_TURSTART, it wasn't terribly useful in the first place. Add a new error action, SS_REQSENSE. (Send a request sense and then retry the command.) This is useful when the controller hasn't performed autosense for some reason. Change the default actions around a bit. scsi_cd.c, scsi_da.c, scsi_pt.c, scsi_ses.c: SF_RETRY_SELTO -> CAM_RETRY_SELTO. Selection timeouts shouldn't be covered by a sense flag. scsi_pass.[ch]: SF_RETRY_SELTO -> CAM_RETRY_SELTO. Get rid of the last vestiges of a read/write interface. libkern/bsearch.c, sys/libkern.h, conf/files: Add bsearch.c, which is needed for some of the new table lookup routines. aic7xxx_freebsd.c: Define AHC_NEW_TRAN_SETTINGS if CAM_NEW_TRAN_CODE is defined. sbuf.h, subr_sbuf.c: Add the appropriate #ifdefs so sbufs can compile and run in userland. Change sbuf_printf() to use vsnprintf() instead of kvprintf(), which is only available in the kernel. Change the source string for sbuf_cpy() and sbuf_cat() to be a const char *. Add __BEGIN_DECLS and __END_DECLS around function prototypes since they're now exported to userland. kdump/mkioctls: Include stdio.h before cam.h since cam.h now includes a function with a FILE * argument. Submitted by: gibbs (mostly) Reviewed by: jdp, marcel (libsbuf makefile changes) Reviewed by: des (sbuf changes) Reviewed by: ken
2001-03-27 05:45:52 +00:00
error = daerror(done_ccb, CAM_RETRY_SELTO, sf);
if (error == ERESTART) {
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
/*
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
* A retry was scheduled, so
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
* just return.
*/
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
cam_periph_unlock(periph);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
return;
}
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
bp = (struct bio *)done_ccb->ccb_h.ccb_bp;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
if (error != 0) {
sys/cam/scsi/scsi_da.c: - Only attempt the closing synchronize cache on a disk if it is still there. - When a device is lost, report the number of outstanding I/Os as they are drained. - When a device is lost, return any unprocessed bios with ENXIO instead of EIO. - Filter asynchronous events, but always allow cam_periph_async() to see them too. Sponsored by: Spectra Logic Corporation
2011-06-14 16:05:00 +00:00
int queued_error;
/*
* return all queued I/O with EIO, so that
* the client can retry these I/Os in the
* proper order should it attempt to recover.
*/
queued_error = EIO;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
sys/cam/scsi/scsi_da.c: - Only attempt the closing synchronize cache on a disk if it is still there. - When a device is lost, report the number of outstanding I/Os as they are drained. - When a device is lost, return any unprocessed bios with ENXIO instead of EIO. - Filter asynchronous events, but always allow cam_periph_async() to see them too. Sponsored by: Spectra Logic Corporation
2011-06-14 16:05:00 +00:00
if (error == ENXIO
&& (softc->flags & DA_FLAG_PACK_INVALID)== 0) {
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
/*
* Catastrophic error. Mark our pack as
* invalid.
*/
Add an xpt_print function to reduce most of the xpt_print_path/printf pairs. Convert the core code to use it. Reviewed by: scsi (various folks weighing in with good advice) MFC after: 2 weeks
2006-12-05 07:45:28 +00:00
/*
* XXX See if this is really a media
* XXX change first?
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
*/
Add an xpt_print function to reduce most of the xpt_print_path/printf pairs. Convert the core code to use it. Reviewed by: scsi (various folks weighing in with good advice) MFC after: 2 weeks
2006-12-05 07:45:28 +00:00
xpt_print(periph->path,
"Invalidating pack\n");
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
softc->flags |= DA_FLAG_PACK_INVALID;
sys/cam/scsi/scsi_da.c: - Only attempt the closing synchronize cache on a disk if it is still there. - When a device is lost, report the number of outstanding I/Os as they are drained. - When a device is lost, return any unprocessed bios with ENXIO instead of EIO. - Filter asynchronous events, but always allow cam_periph_async() to see them too. Sponsored by: Spectra Logic Corporation
2011-06-14 16:05:00 +00:00
queued_error = ENXIO;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
sys/cam/scsi/scsi_da.c: - Only attempt the closing synchronize cache on a disk if it is still there. - When a device is lost, report the number of outstanding I/Os as they are drained. - When a device is lost, return any unprocessed bios with ENXIO instead of EIO. - Filter asynchronous events, but always allow cam_periph_async() to see them too. Sponsored by: Spectra Logic Corporation
2011-06-14 16:05:00 +00:00
bioq_flush(&softc->bio_queue, NULL,
queued_error);
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
if (bp != NULL) {
bp->bio_error = error;
bp->bio_resid = bp->bio_bcount;
bp->bio_flags |= BIO_ERROR;
}
} else if (bp != NULL) {
Fix returning incorrect bio_resid value with failed BIO_DELETE requests. Neither residual length reported for ATA/SCSI command nor one from another BIO_DELETE request are in any way related to the value to be returned.
2013-07-28 19:56:08 +00:00
if (state == DA_CCB_DELETE)
bp->bio_resid = 0;
else
bp->bio_resid = csio->resid;
Complete the bio/buf divorce for all code below devfs::strategy Exceptions: Vinum untouched. This means that it cannot be compiled. Greg Lehey is on the case. CCD not converted yet, casts to struct buf (still safe) atapi-cd casts to struct buf to examine B_PHYS
2000-04-15 05:54:02 +00:00
bp->bio_error = 0;
Remove unused cmd6workaround() calls. MFC after: 1 day
2003-08-06 14:24:16 +00:00
if (bp->bio_resid != 0)
Complete the bio/buf divorce for all code below devfs::strategy Exceptions: Vinum untouched. This means that it cannot be compiled. Greg Lehey is on the case. CCD not converted yet, casts to struct buf (still safe) atapi-cd casts to struct buf to examine B_PHYS
2000-04-15 05:54:02 +00:00
bp->bio_flags |= BIO_ERROR;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
if ((done_ccb->ccb_h.status & CAM_DEV_QFRZN) != 0)
cam_release_devq(done_ccb->ccb_h.path,
/*relsim_flags*/0,
/*reduction*/0,
/*timeout*/0,
/*getcount_only*/0);
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
} else if (bp != NULL) {
Rewrite of the CAM error recovery code. Some of the major changes include: - The SCSI error handling portion of cam_periph_error() has been broken out into a number of subfunctions to better modularize the code that handles the hierarchy of SCSI errors. As a result, the code is now much easier to read. - String handling and error printing has been significantly revamped. We now use sbufs to do string formatting instead of using printfs (for the kernel) and snprintf/strncat (for userland) as before. There is a new catchall error printing routine, cam_error_print() and its string-based counterpart, cam_error_string() that allow the kernel and userland applications to pass in a CCB and have errors printed out properly, whether or not they're SCSI errors. Among other things, this helped eliminate a fair amount of duplicate code in camcontrol. We now print out more information than before, including the CAM status and SCSI status and the error recovery action taken to remedy the problem. - sbufs are now available in userland, via libsbuf. This change was necessary since most of the error printing code is shared between libcam and the kernel. - A new transfer settings interface is included in this checkin. This code is #ifdef'ed out, and is primarily intended to aid discussion with HBA driver authors on the final form the interface should take. There is example code in the ahc(4) driver that implements the HBA driver side of the new interface. The new transfer settings code won't be enabled until we're ready to switch all HBA drivers over to the new interface. src/Makefile.inc1, lib/Makefile: Add libsbuf. It must be built before libcam, since libcam uses sbuf routines. libcam/Makefile: libcam now depends on libsbuf. libsbuf/Makefile: Add a makefile for libsbuf. This pulls in the sbuf sources from sys/kern. bsd.libnames.mk: Add LIBSBUF. camcontrol/Makefile: Add -lsbuf. Since camcontrol is statically linked, we can't depend on the dynamic linker to pull in libsbuf. camcontrol.c: Use cam_error_print() instead of checking for CAM_SCSI_STATUS_ERROR on every failed CCB. sbuf.9: Change the prototypes for sbuf_cat() and sbuf_cpy() so that the source string is now a const char *. This is more in line wth the standard system string functions, and helps eliminate warnings when dealing with a const source buffer. Fix a typo. cam.c: Add description strings for the various CAM error status values, as well as routines to look up those strings. Add new cam_error_string() and cam_error_print() routines for userland and the kernel. cam.h: Add a new CAM flag, CAM_RETRY_SELTO. Add enumerated types for the various options available with cam_error_print() and cam_error_string(). cam_ccb.h: Add new transfer negotiation structures/types. Change inq_len in the ccb_getdev structure to be "reserved". This field has never been filled in, and will be removed when we next bump the CAM version. cam_debug.h: Fix typo. cam_periph.c: Modularize cam_periph_error(). The SCSI error handling part of cam_periph_error() is now in camperiphscsistatuserror() and camperiphscsisenseerror(). In cam_periph_lock(), increase the reference count on the periph while we wait for our lock attempt to succeed so that the periph won't go away while we're sleeping. cam_xpt.c: Add new transfer negotiation code. (ifdefed out) Add a new function, xpt_path_string(). This is a string/sbuf analog to xpt_print_path(). scsi_all.c: Revamp string handing and error printing code. We now use sbufs for much of the string formatting code. More of that code is shared between userland the kernel. scsi_all.h: Get rid of SS_TURSTART, it wasn't terribly useful in the first place. Add a new error action, SS_REQSENSE. (Send a request sense and then retry the command.) This is useful when the controller hasn't performed autosense for some reason. Change the default actions around a bit. scsi_cd.c, scsi_da.c, scsi_pt.c, scsi_ses.c: SF_RETRY_SELTO -> CAM_RETRY_SELTO. Selection timeouts shouldn't be covered by a sense flag. scsi_pass.[ch]: SF_RETRY_SELTO -> CAM_RETRY_SELTO. Get rid of the last vestiges of a read/write interface. libkern/bsearch.c, sys/libkern.h, conf/files: Add bsearch.c, which is needed for some of the new table lookup routines. aic7xxx_freebsd.c: Define AHC_NEW_TRAN_SETTINGS if CAM_NEW_TRAN_CODE is defined. sbuf.h, subr_sbuf.c: Add the appropriate #ifdefs so sbufs can compile and run in userland. Change sbuf_printf() to use vsnprintf() instead of kvprintf(), which is only available in the kernel. Change the source string for sbuf_cpy() and sbuf_cat() to be a const char *. Add __BEGIN_DECLS and __END_DECLS around function prototypes since they're now exported to userland. kdump/mkioctls: Include stdio.h before cam.h since cam.h now includes a function with a FILE * argument. Submitted by: gibbs (mostly) Reviewed by: jdp, marcel (libsbuf makefile changes) Reviewed by: des (sbuf changes) Reviewed by: ken
2001-03-27 05:45:52 +00:00
if ((done_ccb->ccb_h.status & CAM_DEV_QFRZN) != 0)
panic("REQ_CMP with QFRZN");
Fix returning incorrect bio_resid value with failed BIO_DELETE requests. Neither residual length reported for ATA/SCSI command nor one from another BIO_DELETE request are in any way related to the value to be returned.
2013-07-28 19:56:08 +00:00
if (state == DA_CCB_DELETE)
bp->bio_resid = 0;
else
bp->bio_resid = csio->resid;
Remove unused cmd6workaround() calls. MFC after: 1 day
2003-08-06 14:24:16 +00:00
if (csio->resid > 0)
Complete the bio/buf divorce for all code below devfs::strategy Exceptions: Vinum untouched. This means that it cannot be compiled. Greg Lehey is on the case. CCD not converted yet, casts to struct buf (still safe) atapi-cd casts to struct buf to examine B_PHYS
2000-04-15 05:54:02 +00:00
bp->bio_flags |= BIO_ERROR;
Fix a race condition in CAM peripheral free handling, locking in the CAM XPT bus traversal code, and a number of other periph level issues. cam_periph.h, cam_periph.c: Modify cam_periph_acquire() to test the CAM_PERIPH_INVALID flag prior to allowing a reference count to be gained on a peripheral. Callers of this function will receive CAM_REQ_CMP_ERR status in the situation of attempting to reference an invalidated periph. This guarantees that a peripheral scheduled for a deferred free will not be accessed during its wait for destruction. Panic during attempts to drop a reference count on a peripheral that already has a zero reference count. In cam_periph_list(), use a local sbuf with SBUF_FIXEDLEN set so that mallocs do not occur while the xpt topology lock is held, regardless of the allocation policy of the passed in sbuf. Add a new routine, cam_periph_release_locked_buses(), that can be called when the caller already holds the CAM topology lock. Add some extra debugging for duplicate peripheral allocations in cam_periph_alloc(). Treat CAM_DEV_NOT_THERE much the same as a selection timeout (AC_LOST_DEVICE is emitted), but forgo retries. cam_xpt.c: Revamp the way the EDT traversal code does locking and reference counting. This was broken, since it assumed that the EDT would not change during traversal, but that assumption is no longer valid. So, to prevent devices from going away while we traverse the EDT, make sure we properly lock everything and hold references on devices that we are using. The two peripheral driver traversal routines should be examined. xptpdperiphtraverse() holds the topology lock for the entire time it runs. xptperiphtraverse() is now locked properly, but only holds the topology lock while it is traversing the list, and not while the traversal function is running. The bus locking code in xptbustraverse() should also be revisited at a later time, since it is complex and should probably be simplified. scsi_da.c: Pay attention to the return value from cam_periph_acquire(). Return 0 always from daclose() even if the disk is now gone. Add some rudimentary error injection support. scsi_sg.c: Fix reference counting in the sg(4) driver. The sg driver was calling cam_periph_release() on close, but never called cam_periph_acquire() (which increments the reference count) on open. The periph code correctly complained that the sg(4) driver was trying to decrement the refcount when it was already 0. Sponsored by: Spectra Logic MFC after: 2 weeks
2012-01-12 00:41:48 +00:00
if (softc->error_inject != 0) {
bp->bio_error = softc->error_inject;
bp->bio_resid = bp->bio_bcount;
bp->bio_flags |= BIO_ERROR;
softc->error_inject = 0;
}
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
LIST_REMOVE(&done_ccb->ccb_h, periph_links.le);
Some microoptimizations for da and ada drivers: - Replace ordered_tag_count counter with single flag; - From da remove outstanding_cmds counter, duplicating pending_ccbs list; - From da_softc remove unused links field.
2013-10-24 14:05:44 +00:00
if (LIST_EMPTY(&softc->pending_ccbs))
softc->flags |= DA_FLAG_WAS_OTAG;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
xpt_release_ccb(done_ccb);
Do not schedule periph for payload/TUR requests if reprobe is in progress to avoid sending extra READ CAPACITY requests by dastart(). Schedule periph again on reprobe completion, or otherwise it may stuck indefinitely long. This should fix USB explore thread hanging on device unplug, waiting for periph destruction. Reported by: hselasky
2013-01-11 16:10:11 +00:00
if (state == DA_CCB_DELETE) {
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
TAILQ_HEAD(, bio) queue;
TAILQ_INIT(&queue);
TAILQ_CONCAT(&queue, &softc->delete_run_queue.queue, bio_queue);
softc->delete_run_queue.insert_point = NULL;
Explain a bit of tricky code dealing with trims and how it prevents starvation. These side effects aren't obvious without extremely careful study, and are important to do just so.
2015-01-13 00:20:35 +00:00
/*
* Normally, the xpt_release_ccb() above would make sure
* that when we have more work to do, that work would
* get kicked off. However, we specifically keep
* delete_running set to 0 before the call above to
* allow other I/O to progress when many BIO_DELETE
* requests are pushed down. We set delete_running to 0
* and call daschedule again so that we don't stall if
* there are no other I/Os pending apart from BIO_DELETEs.
*/
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
softc->delete_running = 0;
daschedule(periph);
cam_periph_unlock(periph);
while ((bp1 = TAILQ_FIRST(&queue)) != NULL) {
TAILQ_REMOVE(&queue, bp1, bio_queue);
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
bp1->bio_error = bp->bio_error;
Fix returning incorrect bio_resid value with failed BIO_DELETE requests. Neither residual length reported for ATA/SCSI command nor one from another BIO_DELETE request are in any way related to the value to be returned.
2013-07-28 19:56:08 +00:00
if (bp->bio_flags & BIO_ERROR) {
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
bp1->bio_flags |= BIO_ERROR;
Fix returning incorrect bio_resid value with failed BIO_DELETE requests. Neither residual length reported for ATA/SCSI command nor one from another BIO_DELETE request are in any way related to the value to be returned.
2013-07-28 19:56:08 +00:00
bp1->bio_resid = bp1->bio_bcount;
} else
bp1->bio_resid = 0;
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
biodone(bp1);
}
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
} else
cam_periph_unlock(periph);
if (bp != NULL)
Add BIO_DELETE support for SCSI Direct Access devices (da). Depending on device capabilities use different methods to implement it. Currently used method can be read/set via kern.cam.da.X.delete_method sysctls. Possible values are: NONE - no provisioning support reported by the device; DISABLE - provisioning support was disabled because of errors; ZERO - use WRITE SAME (10) command to write zeroes; WS10 - use WRITE SAME (10) command with UNMAP bit set; WS16 - use WRITE SAME (16) command with UNMAP bit set; UNMAP - use UNMAP command (equivalent of the ATA DSM TRIM command). The last two methods (UNMAP and WS16) are defined by SBC specification and the UNMAP method is the most advanced one. The rest of methods I've found supported in Linux, and as soon as they were trivial to implement, then why not? Hope they will be useful in some cases. Unluckily I have no devices properly reporting parameters of the logical block provisioning support via respective VPD pages (0xB0 and 0xB2). So all info I have/use now is the flag telling whether logical block provisioning is supported or not. As result, specific methods chosen now by trying different ones in order (UNMAP, WS16, DISABLE) and checking completion status to fallback if needed. I don't expect problems from this, as if something go wrong, it should just disable itself. It may disable even too aggressively if only some command parameter misfit. Unlike Linux, which executes each delete with separate request, I've implemented here the same request aggregation as implemented in ada driver. Tests on SSDs I have show much better results doing it this way: above 8GB/s of the linear delete on Intel SATA SSD on LSI SAS HBA (mps). Reviewed by: silence on scsi@ MFC after: 2 month Sponsored by: iXsystems, Inc.
2012-01-13 10:21:17 +00:00
biodone(bp);
Merge CAM locking changes from the projects/camlock branch to radically reduce lock congestion and improve SMP scalability of the SCSI/ATA stack, preparing the ground for the coming next GEOM direct dispatch support. Replace big per-SIM locks with bunch of smaller ones: - per-LUN locks to protect device and peripheral drivers state; - per-target locks to protect list of LUNs on target; - per-bus locks to protect reference counting; - per-send queue locks to protect queue of CCBs to be sent; - per-done queue locks to protect queue of completed CCBs; - remaining per-SIM locks now protect only HBA driver internals. While holding LUN lock it is allowed (while not recommended for performance reasons) to take SIM lock. The opposite acquisition order is forbidden. All the other locks are leaf locks, that can be taken anywhere, but should not be cascaded. Many functions, such as: xpt_action(), xpt_done(), xpt_async(), xpt_create_path(), etc. are no longer require (but allow) SIM lock to be held. To keep compatibility and solve cases where SIM lock can't be dropped, all xpt_async() calls in addition to xpt_done() calls are queued to completion threads for async processing in clean environment without SIM lock held. Instead of single CAM SWI thread, used for commands completion processing before, use multiple (depending on number of CPUs) threads. Load balanced between them using "hash" of the device B:T:L address. HBA drivers that can drop SIM lock during completion processing and have sufficient number of completion threads to efficiently scale to multiple CPUs can use new function xpt_done_direct() to avoid extra context switch. Make ahci(4) driver to use this mechanism depending on hardware setup. Sponsored by: iXsystems, Inc. MFC after: 2 months
2013-10-21 12:00:26 +00:00
return;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
case DA_CCB_PROBE_RC:
case DA_CCB_PROBE_RC16:
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
{
struct scsi_read_capacity_data *rdcap;
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
struct scsi_read_capacity_data_long *rcaplong;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
char announce_buf[80];
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
int lbp;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
lbp = 0;
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
rdcap = NULL;
rcaplong = NULL;
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
if (state == DA_CCB_PROBE_RC)
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
rdcap =(struct scsi_read_capacity_data *)csio->data_ptr;
else
rcaplong = (struct scsi_read_capacity_data_long *)
csio->data_ptr;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
if ((csio->ccb_h.status & CAM_STATUS_MASK) == CAM_REQ_CMP) {
struct disk_params *dp;
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
uint32_t block_size;
uint64_t maxsector;
Use READ CAPACITY(16) to get information about device physical sectors. As soon as not all devices support READ CAPACITY(16), automatically fall back to READ CAPACITY(10) if CAM_REQ_INVALID or SSD_KEY_ILLEGAL_REQUEST status returned. It also provides first bits of information about Logical Block Provisioning (aka UNMAP/TRIM) support by the device.
2011-12-23 19:12:02 +00:00
u_int lalba; /* Lowest aligned LBA. */
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
if (state == DA_CCB_PROBE_RC) {
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
block_size = scsi_4btoul(rdcap->length);
maxsector = scsi_4btoul(rdcap->addr);
Use READ CAPACITY(16) to get information about device physical sectors. As soon as not all devices support READ CAPACITY(16), automatically fall back to READ CAPACITY(10) if CAM_REQ_INVALID or SSD_KEY_ILLEGAL_REQUEST status returned. It also provides first bits of information about Logical Block Provisioning (aka UNMAP/TRIM) support by the device.
2011-12-23 19:12:02 +00:00
lalba = 0;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
/*
* According to SBC-2, if the standard 10
* byte READ CAPACITY command returns 2^32,
* we should issue the 16 byte version of
* the command, since the device in question
* has more sectors than can be represented
* with the short version of the command.
*/
if (maxsector == 0xffffffff) {
Eliminate the use of M_TEMP.
2007-05-14 21:48:53 +00:00
free(rdcap, M_SCSIDA);
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
xpt_release_ccb(done_ccb);
Update probe flow so that devices with lbp can also disable disksort. Ensure that delete_available is reset so re-probes after a media change, to one with different delete characteristics, will result in the correct methods being flagged as available. Make all ccb state changes use a consistent flow: * free() * xpt_release_ccb() * softc->state = <new state> * xpt_schedule() Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 14:14:57 +00:00
softc->state = DA_STATE_PROBE_RC16;
MFp4: - Reduce code duplication in ATA XPT and PMP driver. - Move PIO size setting from ada driver to ATA XPT. It is XPT business to negotiate transfer details. ada driver is now stateless. - Report PIO size to SIM. It is required for correct PATA SIM operation. - Tune PMP scan timings. It workarounds some problems with SiI. - If reset hapens during PMP initialization - restart it. - Introduce early-initialized periph drivers, which are used during initial scan process. Use it for xpt, probe, aprobe and pmp. It gives pmp chance to finish scan before mountroot and numerate devices in right order.
2009-10-31 10:43:38 +00:00
xpt_schedule(periph, priority);
Fix three problems in large (>2TB) device handling: - Make sure we don't release the READ CAPACITY CCB twice - If we have a device that needs a 16 byte READ CAPACITY command, make sure we call xpt_schedule() so we can get a CCB. - Don't unlock the peripheral until we're fully probed. Many thanks to Julian Elischer for providing hardware and testing this. Tested by: julian
2003-05-03 00:21:40 +00:00
return;
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
}
} else {
block_size = scsi_4btoul(rcaplong->length);
maxsector = scsi_8btou64(rcaplong->addr);
Use READ CAPACITY(16) to get information about device physical sectors. As soon as not all devices support READ CAPACITY(16), automatically fall back to READ CAPACITY(10) if CAM_REQ_INVALID or SSD_KEY_ILLEGAL_REQUEST status returned. It also provides first bits of information about Logical Block Provisioning (aka UNMAP/TRIM) support by the device.
2011-12-23 19:12:02 +00:00
lalba = scsi_2btoul(rcaplong->lalba_lbp);
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
}
Don't allow attachment of disks that could cause GEOM to panic.
2006-09-16 21:21:07 +00:00
/*
* Because GEOM code just will panic us if we
* give them an 'illegal' value we'll avoid that
* here.
*/
Minor refactoring of code block. MFC after: 1 day
2015-01-19 07:29:07 +00:00
if (block_size == 0) {
Allow a block size of zero to mean 512 bytes, which is the most common block size for USB disks. This fixes support for "Action Cam SJ4000". Reviewed by: mav @ MFC after: 1 week
2015-01-08 15:10:25 +00:00
block_size = 512;
Minor refactoring of code block. MFC after: 1 day
2015-01-19 07:29:07 +00:00
if (maxsector == 0)
maxsector = -1;
Tune support for removable media in da driver: - remove DA_FLAG_SAW_MEDIA flag, almost opposite to DA_FLAG_PACK_INVALID, using the last instead. - allow opening device with no media present, reporting zero media size and non-zero sector size, as geom/notes suggests. That allow to read device attributes and potentially do other things, not related to media.
2013-05-03 11:53:06 +00:00
}
Allow a block size of zero to mean 512 bytes, which is the most common block size for USB disks. This fixes support for "Action Cam SJ4000". Reviewed by: mav @ MFC after: 1 week
2015-01-08 15:10:25 +00:00
if (block_size >= MAXPHYS) {
Add an xpt_print function to reduce most of the xpt_print_path/printf pairs. Convert the core code to use it. Reviewed by: scsi (various folks weighing in with good advice) MFC after: 2 weeks
2006-12-05 07:45:28 +00:00
xpt_print(periph->path,
"unsupportable block size %ju\n",
(uintmax_t) block_size);
Don't allow attachment of disks that could cause GEOM to panic.
2006-09-16 21:21:07 +00:00
announce_buf[0] = '\0';
cam_periph_invalidate(periph);
} else {
Add CAM infrastructure to allow reporting when a drive's long read capacity data changes. cam_ccb.h: Add a new advanced information type, CDAI_TYPE_RCAPLONG, for long read capacity data. cam_xpt_internal.h: Add a read capacity data pointer and length to struct cam_ed. cam_xpt.c: Free the read capacity buffer when a device goes away. While we're here, make sure we don't leak memory for other malloced fields in struct cam_ed. scsi_all.c: Update the scsi_read_capacity_16() to take a uint8_t * and a length instead of just a pointer to the parameter data structure. This will hopefully make this function somewhat immune to future changes in the parameter data. scsi_all.h: Add some extra bit definitions to struct scsi_read_capacity_data_long, and bump up the structure size to the full size specified by SBC-3. Change the prototype for scsi_read_capacity_16(). scsi_da.c: Register changes in read capacity data with the transport layer. This allows the transport layer to send out an async notification to interested parties. Update the dasetgeom() API. Use scsi_extract_sense_len() instead of scsi_extract_sense(). scsi_xpt.c: Add support for the new CDAI_TYPE_RCAPLONG advanced information type. Make sure we set the physpath pointer to NULL after freeing it. This allows blindly freeing it in the struct cam_ed destructor. sys/param.h: Bump __FreeBSD_version from 1000005 to 1000006 to make it easier for third party drivers to determine that the read capacity data async notification is available. camcontrol.c, mptutil/mpt_cam.c: Update these for the new scsi_read_capacity_16() argument structure. Sponsored by: Spectra Logic
2012-01-26 18:09:28 +00:00
/*
* We pass rcaplong into dasetgeom(),
* because it will only use it if it is
* non-NULL.
*/
Use READ CAPACITY(16) to get information about device physical sectors. As soon as not all devices support READ CAPACITY(16), automatically fall back to READ CAPACITY(10) if CAM_REQ_INVALID or SSD_KEY_ILLEGAL_REQUEST status returned. It also provides first bits of information about Logical Block Provisioning (aka UNMAP/TRIM) support by the device.
2011-12-23 19:12:02 +00:00
dasetgeom(periph, block_size, maxsector,
Add CAM infrastructure to allow reporting when a drive's long read capacity data changes. cam_ccb.h: Add a new advanced information type, CDAI_TYPE_RCAPLONG, for long read capacity data. cam_xpt_internal.h: Add a read capacity data pointer and length to struct cam_ed. cam_xpt.c: Free the read capacity buffer when a device goes away. While we're here, make sure we don't leak memory for other malloced fields in struct cam_ed. scsi_all.c: Update the scsi_read_capacity_16() to take a uint8_t * and a length instead of just a pointer to the parameter data structure. This will hopefully make this function somewhat immune to future changes in the parameter data. scsi_all.h: Add some extra bit definitions to struct scsi_read_capacity_data_long, and bump up the structure size to the full size specified by SBC-3. Change the prototype for scsi_read_capacity_16(). scsi_da.c: Register changes in read capacity data with the transport layer. This allows the transport layer to send out an async notification to interested parties. Update the dasetgeom() API. Use scsi_extract_sense_len() instead of scsi_extract_sense(). scsi_xpt.c: Add support for the new CDAI_TYPE_RCAPLONG advanced information type. Make sure we set the physpath pointer to NULL after freeing it. This allows blindly freeing it in the struct cam_ed destructor. sys/param.h: Bump __FreeBSD_version from 1000005 to 1000006 to make it easier for third party drivers to determine that the read capacity data async notification is available. camcontrol.c, mptutil/mpt_cam.c: Update these for the new scsi_read_capacity_16() argument structure. Sponsored by: Spectra Logic
2012-01-26 18:09:28 +00:00
rcaplong, sizeof(*rcaplong));
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
lbp = (lalba & SRC16_LBPME_A);
Don't allow attachment of disks that could cause GEOM to panic.
2006-09-16 21:21:07 +00:00
dp = &softc->params;
snprintf(announce_buf, sizeof(announce_buf),
Remove legacy CHS geometry from dmesg and unify capacity outputs.
2015-10-11 13:48:20 +00:00
"%juMB (%ju %u byte sectors)",
((uintmax_t)dp->secsize * dp->sectors) /
(1024 * 1024),
(uintmax_t)dp->sectors, dp->secsize);
Don't allow attachment of disks that could cause GEOM to panic.
2006-09-16 21:21:07 +00:00
}
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
} else {
int error;
Move initialization of announce_buf up to make sure it is initialized in all cases.
1999-08-15 23:34:40 +00:00
announce_buf[0] = '\0';
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
/*
* Retry any UNIT ATTENTION type errors. They
* are expected at boot.
*/
Rewrite of the CAM error recovery code. Some of the major changes include: - The SCSI error handling portion of cam_periph_error() has been broken out into a number of subfunctions to better modularize the code that handles the hierarchy of SCSI errors. As a result, the code is now much easier to read. - String handling and error printing has been significantly revamped. We now use sbufs to do string formatting instead of using printfs (for the kernel) and snprintf/strncat (for userland) as before. There is a new catchall error printing routine, cam_error_print() and its string-based counterpart, cam_error_string() that allow the kernel and userland applications to pass in a CCB and have errors printed out properly, whether or not they're SCSI errors. Among other things, this helped eliminate a fair amount of duplicate code in camcontrol. We now print out more information than before, including the CAM status and SCSI status and the error recovery action taken to remedy the problem. - sbufs are now available in userland, via libsbuf. This change was necessary since most of the error printing code is shared between libcam and the kernel. - A new transfer settings interface is included in this checkin. This code is #ifdef'ed out, and is primarily intended to aid discussion with HBA driver authors on the final form the interface should take. There is example code in the ahc(4) driver that implements the HBA driver side of the new interface. The new transfer settings code won't be enabled until we're ready to switch all HBA drivers over to the new interface. src/Makefile.inc1, lib/Makefile: Add libsbuf. It must be built before libcam, since libcam uses sbuf routines. libcam/Makefile: libcam now depends on libsbuf. libsbuf/Makefile: Add a makefile for libsbuf. This pulls in the sbuf sources from sys/kern. bsd.libnames.mk: Add LIBSBUF. camcontrol/Makefile: Add -lsbuf. Since camcontrol is statically linked, we can't depend on the dynamic linker to pull in libsbuf. camcontrol.c: Use cam_error_print() instead of checking for CAM_SCSI_STATUS_ERROR on every failed CCB. sbuf.9: Change the prototypes for sbuf_cat() and sbuf_cpy() so that the source string is now a const char *. This is more in line wth the standard system string functions, and helps eliminate warnings when dealing with a const source buffer. Fix a typo. cam.c: Add description strings for the various CAM error status values, as well as routines to look up those strings. Add new cam_error_string() and cam_error_print() routines for userland and the kernel. cam.h: Add a new CAM flag, CAM_RETRY_SELTO. Add enumerated types for the various options available with cam_error_print() and cam_error_string(). cam_ccb.h: Add new transfer negotiation structures/types. Change inq_len in the ccb_getdev structure to be "reserved". This field has never been filled in, and will be removed when we next bump the CAM version. cam_debug.h: Fix typo. cam_periph.c: Modularize cam_periph_error(). The SCSI error handling part of cam_periph_error() is now in camperiphscsistatuserror() and camperiphscsisenseerror(). In cam_periph_lock(), increase the reference count on the periph while we wait for our lock attempt to succeed so that the periph won't go away while we're sleeping. cam_xpt.c: Add new transfer negotiation code. (ifdefed out) Add a new function, xpt_path_string(). This is a string/sbuf analog to xpt_print_path(). scsi_all.c: Revamp string handing and error printing code. We now use sbufs for much of the string formatting code. More of that code is shared between userland the kernel. scsi_all.h: Get rid of SS_TURSTART, it wasn't terribly useful in the first place. Add a new error action, SS_REQSENSE. (Send a request sense and then retry the command.) This is useful when the controller hasn't performed autosense for some reason. Change the default actions around a bit. scsi_cd.c, scsi_da.c, scsi_pt.c, scsi_ses.c: SF_RETRY_SELTO -> CAM_RETRY_SELTO. Selection timeouts shouldn't be covered by a sense flag. scsi_pass.[ch]: SF_RETRY_SELTO -> CAM_RETRY_SELTO. Get rid of the last vestiges of a read/write interface. libkern/bsearch.c, sys/libkern.h, conf/files: Add bsearch.c, which is needed for some of the new table lookup routines. aic7xxx_freebsd.c: Define AHC_NEW_TRAN_SETTINGS if CAM_NEW_TRAN_CODE is defined. sbuf.h, subr_sbuf.c: Add the appropriate #ifdefs so sbufs can compile and run in userland. Change sbuf_printf() to use vsnprintf() instead of kvprintf(), which is only available in the kernel. Change the source string for sbuf_cpy() and sbuf_cat() to be a const char *. Add __BEGIN_DECLS and __END_DECLS around function prototypes since they're now exported to userland. kdump/mkioctls: Include stdio.h before cam.h since cam.h now includes a function with a FILE * argument. Submitted by: gibbs (mostly) Reviewed by: jdp, marcel (libsbuf makefile changes) Reviewed by: des (sbuf changes) Reviewed by: ken
2001-03-27 05:45:52 +00:00
error = daerror(done_ccb, CAM_RETRY_SELTO,
SF_RETRY_UA|SF_NO_PRINT);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
if (error == ERESTART) {
/*
* A retry was scheuled, so
* just return.
*/
return;
} else if (error != 0) {
int asc, ascq;
int sense_key, error_code;
int have_sense;
cam_status status;
struct ccb_getdev cgd;
/* Don't wedge this device's queue */
status = done_ccb->ccb_h.status;
Rewrite of the CAM error recovery code. Some of the major changes include: - The SCSI error handling portion of cam_periph_error() has been broken out into a number of subfunctions to better modularize the code that handles the hierarchy of SCSI errors. As a result, the code is now much easier to read. - String handling and error printing has been significantly revamped. We now use sbufs to do string formatting instead of using printfs (for the kernel) and snprintf/strncat (for userland) as before. There is a new catchall error printing routine, cam_error_print() and its string-based counterpart, cam_error_string() that allow the kernel and userland applications to pass in a CCB and have errors printed out properly, whether or not they're SCSI errors. Among other things, this helped eliminate a fair amount of duplicate code in camcontrol. We now print out more information than before, including the CAM status and SCSI status and the error recovery action taken to remedy the problem. - sbufs are now available in userland, via libsbuf. This change was necessary since most of the error printing code is shared between libcam and the kernel. - A new transfer settings interface is included in this checkin. This code is #ifdef'ed out, and is primarily intended to aid discussion with HBA driver authors on the final form the interface should take. There is example code in the ahc(4) driver that implements the HBA driver side of the new interface. The new transfer settings code won't be enabled until we're ready to switch all HBA drivers over to the new interface. src/Makefile.inc1, lib/Makefile: Add libsbuf. It must be built before libcam, since libcam uses sbuf routines. libcam/Makefile: libcam now depends on libsbuf. libsbuf/Makefile: Add a makefile for libsbuf. This pulls in the sbuf sources from sys/kern. bsd.libnames.mk: Add LIBSBUF. camcontrol/Makefile: Add -lsbuf. Since camcontrol is statically linked, we can't depend on the dynamic linker to pull in libsbuf. camcontrol.c: Use cam_error_print() instead of checking for CAM_SCSI_STATUS_ERROR on every failed CCB. sbuf.9: Change the prototypes for sbuf_cat() and sbuf_cpy() so that the source string is now a const char *. This is more in line wth the standard system string functions, and helps eliminate warnings when dealing with a const source buffer. Fix a typo. cam.c: Add description strings for the various CAM error status values, as well as routines to look up those strings. Add new cam_error_string() and cam_error_print() routines for userland and the kernel. cam.h: Add a new CAM flag, CAM_RETRY_SELTO. Add enumerated types for the various options available with cam_error_print() and cam_error_string(). cam_ccb.h: Add new transfer negotiation structures/types. Change inq_len in the ccb_getdev structure to be "reserved". This field has never been filled in, and will be removed when we next bump the CAM version. cam_debug.h: Fix typo. cam_periph.c: Modularize cam_periph_error(). The SCSI error handling part of cam_periph_error() is now in camperiphscsistatuserror() and camperiphscsisenseerror(). In cam_periph_lock(), increase the reference count on the periph while we wait for our lock attempt to succeed so that the periph won't go away while we're sleeping. cam_xpt.c: Add new transfer negotiation code. (ifdefed out) Add a new function, xpt_path_string(). This is a string/sbuf analog to xpt_print_path(). scsi_all.c: Revamp string handing and error printing code. We now use sbufs for much of the string formatting code. More of that code is shared between userland the kernel. scsi_all.h: Get rid of SS_TURSTART, it wasn't terribly useful in the first place. Add a new error action, SS_REQSENSE. (Send a request sense and then retry the command.) This is useful when the controller hasn't performed autosense for some reason. Change the default actions around a bit. scsi_cd.c, scsi_da.c, scsi_pt.c, scsi_ses.c: SF_RETRY_SELTO -> CAM_RETRY_SELTO. Selection timeouts shouldn't be covered by a sense flag. scsi_pass.[ch]: SF_RETRY_SELTO -> CAM_RETRY_SELTO. Get rid of the last vestiges of a read/write interface. libkern/bsearch.c, sys/libkern.h, conf/files: Add bsearch.c, which is needed for some of the new table lookup routines. aic7xxx_freebsd.c: Define AHC_NEW_TRAN_SETTINGS if CAM_NEW_TRAN_CODE is defined. sbuf.h, subr_sbuf.c: Add the appropriate #ifdefs so sbufs can compile and run in userland. Change sbuf_printf() to use vsnprintf() instead of kvprintf(), which is only available in the kernel. Change the source string for sbuf_cpy() and sbuf_cat() to be a const char *. Add __BEGIN_DECLS and __END_DECLS around function prototypes since they're now exported to userland. kdump/mkioctls: Include stdio.h before cam.h since cam.h now includes a function with a FILE * argument. Submitted by: gibbs (mostly) Reviewed by: jdp, marcel (libsbuf makefile changes) Reviewed by: des (sbuf changes) Reviewed by: ken
2001-03-27 05:45:52 +00:00
if ((status & CAM_DEV_QFRZN) != 0)
cam_release_devq(done_ccb->ccb_h.path,
/*relsim_flags*/0,
/*reduction*/0,
/*timeout*/0,
/*getcount_only*/0);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
xpt_setup_ccb(&cgd.ccb_h,
done_ccb->ccb_h.path,
Replace most of priority numbers with defines. No logical changes.
2009-10-23 08:27:55 +00:00
CAM_PRIORITY_NORMAL);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
cgd.ccb_h.func_code = XPT_GDEV_TYPE;
xpt_action((union ccb *)&cgd);
Add scsi_extract_sense_ccb() -- wrapper around scsi_extract_sense_len(). It allows to remove number of duplicate checks from several places.
2012-06-23 12:32:53 +00:00
if (scsi_extract_sense_ccb(done_ccb,
&error_code, &sense_key, &asc, &ascq))
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
have_sense = TRUE;
Add scsi_extract_sense_ccb() -- wrapper around scsi_extract_sense_len(). It allows to remove number of duplicate checks from several places.
2012-06-23 12:32:53 +00:00
else
have_sense = FALSE;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
Use READ CAPACITY(16) to get information about device physical sectors. As soon as not all devices support READ CAPACITY(16), automatically fall back to READ CAPACITY(10) if CAM_REQ_INVALID or SSD_KEY_ILLEGAL_REQUEST status returned. It also provides first bits of information about Logical Block Provisioning (aka UNMAP/TRIM) support by the device.
2011-12-23 19:12:02 +00:00
/*
* If we tried READ CAPACITY(16) and failed,
* fallback to READ CAPACITY(10).
*/
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
if ((state == DA_CCB_PROBE_RC16) &&
Use READ CAPACITY(16) to get information about device physical sectors. As soon as not all devices support READ CAPACITY(16), automatically fall back to READ CAPACITY(10) if CAM_REQ_INVALID or SSD_KEY_ILLEGAL_REQUEST status returned. It also provides first bits of information about Logical Block Provisioning (aka UNMAP/TRIM) support by the device.
2011-12-23 19:12:02 +00:00
(softc->flags & DA_FLAG_CAN_RC16) &&
(((csio->ccb_h.status & CAM_STATUS_MASK) ==
CAM_REQ_INVALID) ||
((have_sense) &&
(error_code == SSD_CURRENT_ERROR) &&
(sense_key == SSD_KEY_ILLEGAL_REQUEST)))) {
softc->flags &= ~DA_FLAG_CAN_RC16;
free(rdcap, M_SCSIDA);
xpt_release_ccb(done_ccb);
Update probe flow so that devices with lbp can also disable disksort. Ensure that delete_available is reset so re-probes after a media change, to one with different delete characteristics, will result in the correct methods being flagged as available. Make all ccb state changes use a consistent flow: * free() * xpt_release_ccb() * softc->state = <new state> * xpt_schedule() Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 14:14:57 +00:00
softc->state = DA_STATE_PROBE_RC;
Use READ CAPACITY(16) to get information about device physical sectors. As soon as not all devices support READ CAPACITY(16), automatically fall back to READ CAPACITY(10) if CAM_REQ_INVALID or SSD_KEY_ILLEGAL_REQUEST status returned. It also provides first bits of information about Logical Block Provisioning (aka UNMAP/TRIM) support by the device.
2011-12-23 19:12:02 +00:00
xpt_schedule(periph, priority);
return;
Remove a stray else. It isn't needed (due to the return at the end of the if statement it pairs with). While not an error today, a careless edit in the future could cause problems (though given the nature of this specific code, the problems quite likely would be some variation of "most direct access SCSI storage devices won't attach," which is unlikely to go unnoticed). PVS-Studio: V705
2016-02-18 15:12:52 +00:00
}
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
/*
Add a number of interrelated CAM feature enhancements and bug fixes. NOTE: These changes will require recompilation of any userland applications, like cdrecord, xmcd, etc., that use the CAM passthrough interface. A make world is recommended. camcontrol.[c8]: - We now support two new commands, "tags" and "negotiate". - The tags commands allows users to view the number of tagged openings for a device as well as a number of other related parameters, and it allows users to set tagged openings for a device. - The negotiate command allows users to enable and disable disconnection and tagged queueing, set sync rates, offsets and bus width. Note that not all of those features are available for all controllers. Only the adv, ahc, and ncr drivers fully support all of the features at this point. Some cards do not allow the setting of sync rates, offsets and the like, and some of the drivers don't have any facilities to do so. Some drivers, like the adw driver, only support enabling or disabling sync negotiation, but do not support setting sync rates. - new description in the camcontrol man page of how to format a disk - cleanup of the camcontrol inquiry command - add support in the 'devlist' command for skipping unconfigured devices if -v was not specified on the command line. - make use of the new base_transfer_speed in the path inquiry CCB. - fix CCB bzero cases cam_xpt.c, cam_sim.[ch], cam_ccb.h: - new flags on many CCB function codes to designate whether they're non-immediate, use a user-supplied CCB, and can only be passed from userland programs via the xpt device. Use these flags in the transport layer and pass driver to categorize CCBs. - new flag in the transport layer device matching code for device nodes that indicates whether a device is unconfigured - bump the CAM version from 0x10 to 0x11 - Change the CAM ioctls to use the version as their group code, so we can force users to recompile code even when the CCB size doesn't change. - add + fill in a new value in the path inquiry CCB, base_transfer_speed. Remove a corresponding field from the cam_sim structure, and add code to every SIM to set this field to the proper value. - Fix the set transfer settings code in the transport layer. scsi_cd.c: - make some variables volatile instead of just casting them in various places - fix a race condition in the changer code - attach unless we get a "logical unit not supported" error. This should fix all of the cases where people have devices that return weird errors when they don't have media in the drive. scsi_da.c: - attach unless we get a "logical unit not supported" error scsi_pass.c: - for immediate CCBs, just malloc a CCB to send the user request in. This gets rid of the 'held' count problem in camcontrol tags. scsi_pass.h: - change the CAM ioctls to use the CAM version as their group code. adv driver: - Allow changing the sync rate and offset separately. adw driver - Allow changing the sync rate and offset separately. aha driver: - Don't return CAM_REQ_CMP for SET_TRAN_SETTINGS CCBs. ahc driver: - Allow setting offset and sync rate separately bt driver: - Don't return CAM_REQ_CMP for SET_TRAN_SETTINGS CCBs. NCR driver: - Fix the ultra/ultra 2 negotiation bug - allow setting both the sync rate and offset separately Other HBA drivers: - Put code in to set the base_transfer_speed field for XPT_GET_TRAN_SETTINGS CCBs. Reviewed by: gibbs, mjacob (isp), imp (aha)
1999-05-06 20:16:39 +00:00
* Attach to anything that claims to be a
* direct access or optical disk device,
* as long as it doesn't return a "Logical
* unit not supported" (0x25) error.
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
*/
Add a number of interrelated CAM feature enhancements and bug fixes. NOTE: These changes will require recompilation of any userland applications, like cdrecord, xmcd, etc., that use the CAM passthrough interface. A make world is recommended. camcontrol.[c8]: - We now support two new commands, "tags" and "negotiate". - The tags commands allows users to view the number of tagged openings for a device as well as a number of other related parameters, and it allows users to set tagged openings for a device. - The negotiate command allows users to enable and disable disconnection and tagged queueing, set sync rates, offsets and bus width. Note that not all of those features are available for all controllers. Only the adv, ahc, and ncr drivers fully support all of the features at this point. Some cards do not allow the setting of sync rates, offsets and the like, and some of the drivers don't have any facilities to do so. Some drivers, like the adw driver, only support enabling or disabling sync negotiation, but do not support setting sync rates. - new description in the camcontrol man page of how to format a disk - cleanup of the camcontrol inquiry command - add support in the 'devlist' command for skipping unconfigured devices if -v was not specified on the command line. - make use of the new base_transfer_speed in the path inquiry CCB. - fix CCB bzero cases cam_xpt.c, cam_sim.[ch], cam_ccb.h: - new flags on many CCB function codes to designate whether they're non-immediate, use a user-supplied CCB, and can only be passed from userland programs via the xpt device. Use these flags in the transport layer and pass driver to categorize CCBs. - new flag in the transport layer device matching code for device nodes that indicates whether a device is unconfigured - bump the CAM version from 0x10 to 0x11 - Change the CAM ioctls to use the version as their group code, so we can force users to recompile code even when the CCB size doesn't change. - add + fill in a new value in the path inquiry CCB, base_transfer_speed. Remove a corresponding field from the cam_sim structure, and add code to every SIM to set this field to the proper value. - Fix the set transfer settings code in the transport layer. scsi_cd.c: - make some variables volatile instead of just casting them in various places - fix a race condition in the changer code - attach unless we get a "logical unit not supported" error. This should fix all of the cases where people have devices that return weird errors when they don't have media in the drive. scsi_da.c: - attach unless we get a "logical unit not supported" error scsi_pass.c: - for immediate CCBs, just malloc a CCB to send the user request in. This gets rid of the 'held' count problem in camcontrol tags. scsi_pass.h: - change the CAM ioctls to use the CAM version as their group code. adv driver: - Allow changing the sync rate and offset separately. adw driver - Allow changing the sync rate and offset separately. aha driver: - Don't return CAM_REQ_CMP for SET_TRAN_SETTINGS CCBs. ahc driver: - Allow setting offset and sync rate separately bt driver: - Don't return CAM_REQ_CMP for SET_TRAN_SETTINGS CCBs. NCR driver: - Fix the ultra/ultra 2 negotiation bug - allow setting both the sync rate and offset separately Other HBA drivers: - Put code in to set the base_transfer_speed field for XPT_GET_TRAN_SETTINGS CCBs. Reviewed by: gibbs, mjacob (isp), imp (aha)
1999-05-06 20:16:39 +00:00
if ((have_sense) && (asc != 0x25)
Rewrite of the CAM error recovery code. Some of the major changes include: - The SCSI error handling portion of cam_periph_error() has been broken out into a number of subfunctions to better modularize the code that handles the hierarchy of SCSI errors. As a result, the code is now much easier to read. - String handling and error printing has been significantly revamped. We now use sbufs to do string formatting instead of using printfs (for the kernel) and snprintf/strncat (for userland) as before. There is a new catchall error printing routine, cam_error_print() and its string-based counterpart, cam_error_string() that allow the kernel and userland applications to pass in a CCB and have errors printed out properly, whether or not they're SCSI errors. Among other things, this helped eliminate a fair amount of duplicate code in camcontrol. We now print out more information than before, including the CAM status and SCSI status and the error recovery action taken to remedy the problem. - sbufs are now available in userland, via libsbuf. This change was necessary since most of the error printing code is shared between libcam and the kernel. - A new transfer settings interface is included in this checkin. This code is #ifdef'ed out, and is primarily intended to aid discussion with HBA driver authors on the final form the interface should take. There is example code in the ahc(4) driver that implements the HBA driver side of the new interface. The new transfer settings code won't be enabled until we're ready to switch all HBA drivers over to the new interface. src/Makefile.inc1, lib/Makefile: Add libsbuf. It must be built before libcam, since libcam uses sbuf routines. libcam/Makefile: libcam now depends on libsbuf. libsbuf/Makefile: Add a makefile for libsbuf. This pulls in the sbuf sources from sys/kern. bsd.libnames.mk: Add LIBSBUF. camcontrol/Makefile: Add -lsbuf. Since camcontrol is statically linked, we can't depend on the dynamic linker to pull in libsbuf. camcontrol.c: Use cam_error_print() instead of checking for CAM_SCSI_STATUS_ERROR on every failed CCB. sbuf.9: Change the prototypes for sbuf_cat() and sbuf_cpy() so that the source string is now a const char *. This is more in line wth the standard system string functions, and helps eliminate warnings when dealing with a const source buffer. Fix a typo. cam.c: Add description strings for the various CAM error status values, as well as routines to look up those strings. Add new cam_error_string() and cam_error_print() routines for userland and the kernel. cam.h: Add a new CAM flag, CAM_RETRY_SELTO. Add enumerated types for the various options available with cam_error_print() and cam_error_string(). cam_ccb.h: Add new transfer negotiation structures/types. Change inq_len in the ccb_getdev structure to be "reserved". This field has never been filled in, and will be removed when we next bump the CAM version. cam_debug.h: Fix typo. cam_periph.c: Modularize cam_periph_error(). The SCSI error handling part of cam_periph_error() is now in camperiphscsistatuserror() and camperiphscsisenseerror(). In cam_periph_lock(), increase the reference count on the periph while we wait for our lock attempt to succeed so that the periph won't go away while we're sleeping. cam_xpt.c: Add new transfer negotiation code. (ifdefed out) Add a new function, xpt_path_string(). This is a string/sbuf analog to xpt_print_path(). scsi_all.c: Revamp string handing and error printing code. We now use sbufs for much of the string formatting code. More of that code is shared between userland the kernel. scsi_all.h: Get rid of SS_TURSTART, it wasn't terribly useful in the first place. Add a new error action, SS_REQSENSE. (Send a request sense and then retry the command.) This is useful when the controller hasn't performed autosense for some reason. Change the default actions around a bit. scsi_cd.c, scsi_da.c, scsi_pt.c, scsi_ses.c: SF_RETRY_SELTO -> CAM_RETRY_SELTO. Selection timeouts shouldn't be covered by a sense flag. scsi_pass.[ch]: SF_RETRY_SELTO -> CAM_RETRY_SELTO. Get rid of the last vestiges of a read/write interface. libkern/bsearch.c, sys/libkern.h, conf/files: Add bsearch.c, which is needed for some of the new table lookup routines. aic7xxx_freebsd.c: Define AHC_NEW_TRAN_SETTINGS if CAM_NEW_TRAN_CODE is defined. sbuf.h, subr_sbuf.c: Add the appropriate #ifdefs so sbufs can compile and run in userland. Change sbuf_printf() to use vsnprintf() instead of kvprintf(), which is only available in the kernel. Change the source string for sbuf_cpy() and sbuf_cat() to be a const char *. Add __BEGIN_DECLS and __END_DECLS around function prototypes since they're now exported to userland. kdump/mkioctls: Include stdio.h before cam.h since cam.h now includes a function with a FILE * argument. Submitted by: gibbs (mostly) Reviewed by: jdp, marcel (libsbuf makefile changes) Reviewed by: des (sbuf changes) Reviewed by: ken
2001-03-27 05:45:52 +00:00
&& (error_code == SSD_CURRENT_ERROR)) {
const char *sense_key_desc;
const char *asc_desc;
Tune support for removable media in da driver: - remove DA_FLAG_SAW_MEDIA flag, almost opposite to DA_FLAG_PACK_INVALID, using the last instead. - allow opening device with no media present, reporting zero media size and non-zero sector size, as geom/notes suggests. That allow to read device attributes and potentially do other things, not related to media.
2013-05-03 11:53:06 +00:00
dasetgeom(periph, 512, -1, NULL, 0);
Rewrite of the CAM error recovery code. Some of the major changes include: - The SCSI error handling portion of cam_periph_error() has been broken out into a number of subfunctions to better modularize the code that handles the hierarchy of SCSI errors. As a result, the code is now much easier to read. - String handling and error printing has been significantly revamped. We now use sbufs to do string formatting instead of using printfs (for the kernel) and snprintf/strncat (for userland) as before. There is a new catchall error printing routine, cam_error_print() and its string-based counterpart, cam_error_string() that allow the kernel and userland applications to pass in a CCB and have errors printed out properly, whether or not they're SCSI errors. Among other things, this helped eliminate a fair amount of duplicate code in camcontrol. We now print out more information than before, including the CAM status and SCSI status and the error recovery action taken to remedy the problem. - sbufs are now available in userland, via libsbuf. This change was necessary since most of the error printing code is shared between libcam and the kernel. - A new transfer settings interface is included in this checkin. This code is #ifdef'ed out, and is primarily intended to aid discussion with HBA driver authors on the final form the interface should take. There is example code in the ahc(4) driver that implements the HBA driver side of the new interface. The new transfer settings code won't be enabled until we're ready to switch all HBA drivers over to the new interface. src/Makefile.inc1, lib/Makefile: Add libsbuf. It must be built before libcam, since libcam uses sbuf routines. libcam/Makefile: libcam now depends on libsbuf. libsbuf/Makefile: Add a makefile for libsbuf. This pulls in the sbuf sources from sys/kern. bsd.libnames.mk: Add LIBSBUF. camcontrol/Makefile: Add -lsbuf. Since camcontrol is statically linked, we can't depend on the dynamic linker to pull in libsbuf. camcontrol.c: Use cam_error_print() instead of checking for CAM_SCSI_STATUS_ERROR on every failed CCB. sbuf.9: Change the prototypes for sbuf_cat() and sbuf_cpy() so that the source string is now a const char *. This is more in line wth the standard system string functions, and helps eliminate warnings when dealing with a const source buffer. Fix a typo. cam.c: Add description strings for the various CAM error status values, as well as routines to look up those strings. Add new cam_error_string() and cam_error_print() routines for userland and the kernel. cam.h: Add a new CAM flag, CAM_RETRY_SELTO. Add enumerated types for the various options available with cam_error_print() and cam_error_string(). cam_ccb.h: Add new transfer negotiation structures/types. Change inq_len in the ccb_getdev structure to be "reserved". This field has never been filled in, and will be removed when we next bump the CAM version. cam_debug.h: Fix typo. cam_periph.c: Modularize cam_periph_error(). The SCSI error handling part of cam_periph_error() is now in camperiphscsistatuserror() and camperiphscsisenseerror(). In cam_periph_lock(), increase the reference count on the periph while we wait for our lock attempt to succeed so that the periph won't go away while we're sleeping. cam_xpt.c: Add new transfer negotiation code. (ifdefed out) Add a new function, xpt_path_string(). This is a string/sbuf analog to xpt_print_path(). scsi_all.c: Revamp string handing and error printing code. We now use sbufs for much of the string formatting code. More of that code is shared between userland the kernel. scsi_all.h: Get rid of SS_TURSTART, it wasn't terribly useful in the first place. Add a new error action, SS_REQSENSE. (Send a request sense and then retry the command.) This is useful when the controller hasn't performed autosense for some reason. Change the default actions around a bit. scsi_cd.c, scsi_da.c, scsi_pt.c, scsi_ses.c: SF_RETRY_SELTO -> CAM_RETRY_SELTO. Selection timeouts shouldn't be covered by a sense flag. scsi_pass.[ch]: SF_RETRY_SELTO -> CAM_RETRY_SELTO. Get rid of the last vestiges of a read/write interface. libkern/bsearch.c, sys/libkern.h, conf/files: Add bsearch.c, which is needed for some of the new table lookup routines. aic7xxx_freebsd.c: Define AHC_NEW_TRAN_SETTINGS if CAM_NEW_TRAN_CODE is defined. sbuf.h, subr_sbuf.c: Add the appropriate #ifdefs so sbufs can compile and run in userland. Change sbuf_printf() to use vsnprintf() instead of kvprintf(), which is only available in the kernel. Change the source string for sbuf_cpy() and sbuf_cat() to be a const char *. Add __BEGIN_DECLS and __END_DECLS around function prototypes since they're now exported to userland. kdump/mkioctls: Include stdio.h before cam.h since cam.h now includes a function with a FILE * argument. Submitted by: gibbs (mostly) Reviewed by: jdp, marcel (libsbuf makefile changes) Reviewed by: des (sbuf changes) Reviewed by: ken
2001-03-27 05:45:52 +00:00
scsi_sense_desc(sense_key, asc, ascq,
&cgd.inq_data,
&sense_key_desc,
&asc_desc);
Examine all occurrences of sprintf(), strcat(), and str[n]cpy() for possible buffer overflow problems. Replaced most sprintf()'s with snprintf(); for others cases, added terminating NUL bytes where appropriate, replaced constants like "16" with sizeof(), etc. These changes include several bug fixes, but most changes are for maintainability's sake. Any instance where it wasn't "immediately obvious" that a buffer overflow could not occur was made safer. Reviewed by: Bruce Evans <bde@zeta.org.au> Reviewed by: Matthew Dillon <dillon@apollo.backplane.com> Reviewed by: Mike Spengler <mks@networkcs.com>
1998-12-04 22:54:57 +00:00
snprintf(announce_buf,
sizeof(announce_buf),
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
"Attempt to query device "
"size failed: %s, %s",
Rewrite of the CAM error recovery code. Some of the major changes include: - The SCSI error handling portion of cam_periph_error() has been broken out into a number of subfunctions to better modularize the code that handles the hierarchy of SCSI errors. As a result, the code is now much easier to read. - String handling and error printing has been significantly revamped. We now use sbufs to do string formatting instead of using printfs (for the kernel) and snprintf/strncat (for userland) as before. There is a new catchall error printing routine, cam_error_print() and its string-based counterpart, cam_error_string() that allow the kernel and userland applications to pass in a CCB and have errors printed out properly, whether or not they're SCSI errors. Among other things, this helped eliminate a fair amount of duplicate code in camcontrol. We now print out more information than before, including the CAM status and SCSI status and the error recovery action taken to remedy the problem. - sbufs are now available in userland, via libsbuf. This change was necessary since most of the error printing code is shared between libcam and the kernel. - A new transfer settings interface is included in this checkin. This code is #ifdef'ed out, and is primarily intended to aid discussion with HBA driver authors on the final form the interface should take. There is example code in the ahc(4) driver that implements the HBA driver side of the new interface. The new transfer settings code won't be enabled until we're ready to switch all HBA drivers over to the new interface. src/Makefile.inc1, lib/Makefile: Add libsbuf. It must be built before libcam, since libcam uses sbuf routines. libcam/Makefile: libcam now depends on libsbuf. libsbuf/Makefile: Add a makefile for libsbuf. This pulls in the sbuf sources from sys/kern. bsd.libnames.mk: Add LIBSBUF. camcontrol/Makefile: Add -lsbuf. Since camcontrol is statically linked, we can't depend on the dynamic linker to pull in libsbuf. camcontrol.c: Use cam_error_print() instead of checking for CAM_SCSI_STATUS_ERROR on every failed CCB. sbuf.9: Change the prototypes for sbuf_cat() and sbuf_cpy() so that the source string is now a const char *. This is more in line wth the standard system string functions, and helps eliminate warnings when dealing with a const source buffer. Fix a typo. cam.c: Add description strings for the various CAM error status values, as well as routines to look up those strings. Add new cam_error_string() and cam_error_print() routines for userland and the kernel. cam.h: Add a new CAM flag, CAM_RETRY_SELTO. Add enumerated types for the various options available with cam_error_print() and cam_error_string(). cam_ccb.h: Add new transfer negotiation structures/types. Change inq_len in the ccb_getdev structure to be "reserved". This field has never been filled in, and will be removed when we next bump the CAM version. cam_debug.h: Fix typo. cam_periph.c: Modularize cam_periph_error(). The SCSI error handling part of cam_periph_error() is now in camperiphscsistatuserror() and camperiphscsisenseerror(). In cam_periph_lock(), increase the reference count on the periph while we wait for our lock attempt to succeed so that the periph won't go away while we're sleeping. cam_xpt.c: Add new transfer negotiation code. (ifdefed out) Add a new function, xpt_path_string(). This is a string/sbuf analog to xpt_print_path(). scsi_all.c: Revamp string handing and error printing code. We now use sbufs for much of the string formatting code. More of that code is shared between userland the kernel. scsi_all.h: Get rid of SS_TURSTART, it wasn't terribly useful in the first place. Add a new error action, SS_REQSENSE. (Send a request sense and then retry the command.) This is useful when the controller hasn't performed autosense for some reason. Change the default actions around a bit. scsi_cd.c, scsi_da.c, scsi_pt.c, scsi_ses.c: SF_RETRY_SELTO -> CAM_RETRY_SELTO. Selection timeouts shouldn't be covered by a sense flag. scsi_pass.[ch]: SF_RETRY_SELTO -> CAM_RETRY_SELTO. Get rid of the last vestiges of a read/write interface. libkern/bsearch.c, sys/libkern.h, conf/files: Add bsearch.c, which is needed for some of the new table lookup routines. aic7xxx_freebsd.c: Define AHC_NEW_TRAN_SETTINGS if CAM_NEW_TRAN_CODE is defined. sbuf.h, subr_sbuf.c: Add the appropriate #ifdefs so sbufs can compile and run in userland. Change sbuf_printf() to use vsnprintf() instead of kvprintf(), which is only available in the kernel. Change the source string for sbuf_cpy() and sbuf_cat() to be a const char *. Add __BEGIN_DECLS and __END_DECLS around function prototypes since they're now exported to userland. kdump/mkioctls: Include stdio.h before cam.h since cam.h now includes a function with a FILE * argument. Submitted by: gibbs (mostly) Reviewed by: jdp, marcel (libsbuf makefile changes) Reviewed by: des (sbuf changes) Reviewed by: ken
2001-03-27 05:45:52 +00:00
sense_key_desc,
asc_desc);
} else {
Some fixes for the CD and DA drivers from bde. (and some of my own as well) Among them: [ cd driver ] 1. Old labeling code was still there. 2. Error handling for dsopen() was broken (no test for the `error' returned by dsopen(); bogus test of an `error' that is known to be 0). 3. cdopen() closed the physical device after certain errors although there may still be open partitions on it. 4. cdclose() closed the physical device although there may still be open partitions on it. 5. Some printf format fixes was incomplete or missing. 6. cdioctl() truncated unit numbers mod 256. 7. cdioctl() was missing locking. [ da driver ] 1. daclose() closed the physical device although there may still be open partitions on it. This was fixed many years ago in sd.c rev.1.57. 2. A minor optimization (the dk_slices != NULL test) in sdopen() became uglier in daopen(). It is not worth doing. da only regressed compared with od and my version of sd, since I never committed the change to sd. daopen() should probably do less if some partition is already open. This is not addressed by the diffs. [ ... ] 5. "opt_hw_wdog.h" was not included, so the HW_WDOG code was unreachable. - Added a getdev CCB call in the cdopen() and daopen() calls so that the vendor name and device name are available for the disklabel. (suggested by bde) - Removed vestigal devfs support in both drivers, since we can't properly work with devfs yet. (ask bde for details on this) - Cleaned up the probe code in both drivers in the failure cases. There were a number of things wrong here. The peripheral driver instances weren't getting properly cleaned up. Sometimes the wrong probe message would get printed out (with the failure message appended), so it wasn't very clear that we failed to attach. SCSI sense information was printed, even when the error in question wasn't a SCSI error. I put similar fixes into the changer driver in revision 1.2 of scsi_ch.c. Reviewed by: gibbs Submitted by: bde (partially)
1998-10-07 02:57:57 +00:00
if (have_sense)
scsi_sense_print(
&done_ccb->csio);
else {
Add an xpt_print function to reduce most of the xpt_print_path/printf pairs. Convert the core code to use it. Reviewed by: scsi (various folks weighing in with good advice) MFC after: 2 weeks
2006-12-05 07:45:28 +00:00
xpt_print(periph->path,
"got CAM status %#x\n",
done_ccb->ccb_h.status);
Some fixes for the CD and DA drivers from bde. (and some of my own as well) Among them: [ cd driver ] 1. Old labeling code was still there. 2. Error handling for dsopen() was broken (no test for the `error' returned by dsopen(); bogus test of an `error' that is known to be 0). 3. cdopen() closed the physical device after certain errors although there may still be open partitions on it. 4. cdclose() closed the physical device although there may still be open partitions on it. 5. Some printf format fixes was incomplete or missing. 6. cdioctl() truncated unit numbers mod 256. 7. cdioctl() was missing locking. [ da driver ] 1. daclose() closed the physical device although there may still be open partitions on it. This was fixed many years ago in sd.c rev.1.57. 2. A minor optimization (the dk_slices != NULL test) in sdopen() became uglier in daopen(). It is not worth doing. da only regressed compared with od and my version of sd, since I never committed the change to sd. daopen() should probably do less if some partition is already open. This is not addressed by the diffs. [ ... ] 5. "opt_hw_wdog.h" was not included, so the HW_WDOG code was unreachable. - Added a getdev CCB call in the cdopen() and daopen() calls so that the vendor name and device name are available for the disklabel. (suggested by bde) - Removed vestigal devfs support in both drivers, since we can't properly work with devfs yet. (ask bde for details on this) - Cleaned up the probe code in both drivers in the failure cases. There were a number of things wrong here. The peripheral driver instances weren't getting properly cleaned up. Sometimes the wrong probe message would get printed out (with the failure message appended), so it wasn't very clear that we failed to attach. SCSI sense information was printed, even when the error in question wasn't a SCSI error. I put similar fixes into the changer driver in revision 1.2 of scsi_ch.c. Reviewed by: gibbs Submitted by: bde (partially)
1998-10-07 02:57:57 +00:00
}
A better fix to avoid race conditions between failed probes and peripheral removal. Obtained from:gibbs@freebsd.org
1999-01-07 20:19:09 +00:00
Add an xpt_print function to reduce most of the xpt_print_path/printf pairs. Convert the core code to use it. Reviewed by: scsi (various folks weighing in with good advice) MFC after: 2 weeks
2006-12-05 07:45:28 +00:00
xpt_print(periph->path, "fatal error, "
"failed to attach to device\n");
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
/*
* Free up resources.
*/
cam_periph_invalidate(periph);
}
}
}
Eliminate the use of M_TEMP.
2007-05-14 21:48:53 +00:00
free(csio->data_ptr, M_SCSIDA);
Do not reread SCSI disk VPD pages on every device open. Instead of rereading VPD pages on every device open, do it only on initial device probe, and in cases when device reported via UNIT ATTENTIONs that something has changed. Capacity is still rereaded on every open because it is more critical for operation and more probable to change in run time. On my tests with Intel 530 SSDs on mps(4) HBA this change reduces time GEOM needs to retaste the device (that includes few open/close cycles) from ~150ms to ~30ms. MFC after: 2 weeks Sponsored by: iXsystems, Inc.
2014-04-30 17:38:26 +00:00
if (announce_buf[0] != '\0' &&
((softc->flags & DA_FLAG_ANNOUNCED) == 0)) {
Move dynamic sysctl(8) variable creation for the cd(4) and da(4) drivers out of cdregister() and daregister(), which are run from interrupt context. The sysctl code does blocking mallocs (M_WAITOK), which causes problems if malloc(9) actually needs to sleep. The eventual fix for this issue will involve moving the CAM probe process inside a kernel thread. For now, though, I have fixed the issue by moving dynamic sysctl variable creation for these two drivers to a task queue running in a kernel thread. The existing task queues (taskqueue_swi and taskqueue_swi_giant) run in software interrupt handlers, which wouldn't fix the problem at hand. So I have created a new task queue, taskqueue_thread, that runs inside a kernel thread. (It also runs outside of Giant -- clients must explicitly acquire and release Giant in their taskqueue functions.) scsi_cd.c: Remove sysctl variable creation code from cdregister(), and move it to a new function, cdsysctlinit(). Queue cdsysctlinit() to the taskqueue_thread taskqueue once we have fully registered the cd(4) driver instance. scsi_da.c: Remove sysctl variable creation code from daregister(), and move it to move it to a new function, dasysctlinit(). Queue dasysctlinit() to the taskqueue_thread taskqueue once we have fully registered the da(4) instance. taskqueue.h: Declare the new taskqueue_thread taskqueue, update some comments. subr_taskqueue.c: Create the new kernel thread taskqueue. This taskqueue runs outside of Giant, so any functions queued to it would need to explicitly acquire/release Giant if they need it. cd.4: Update the cd(4) man page to talk about the minimum command size sysctl/loader tunable. Also note that the changer variables are available as loader tunables as well. da.4: Update the da(4) man page to cover the retry_count, default_timeout and minimum_cmd_size sysctl variables/loader tunables. Remove references to /dev/r???, they aren't used any longer. cd.9: Update the cd(9) man page to describe the CD_Q_10_BYTE_ONLY quirk. taskqueue.9: Update the taskqueue(9) man page to describe the new thread task queue, and the taskqueue_swi_giant queue. MFC after: 3 days
2003-09-03 04:46:28 +00:00
/*
* Create our sysctl variables, now that we know
* we have successfully attached.
*/
Fix a race condition in CAM peripheral free handling, locking in the CAM XPT bus traversal code, and a number of other periph level issues. cam_periph.h, cam_periph.c: Modify cam_periph_acquire() to test the CAM_PERIPH_INVALID flag prior to allowing a reference count to be gained on a peripheral. Callers of this function will receive CAM_REQ_CMP_ERR status in the situation of attempting to reference an invalidated periph. This guarantees that a peripheral scheduled for a deferred free will not be accessed during its wait for destruction. Panic during attempts to drop a reference count on a peripheral that already has a zero reference count. In cam_periph_list(), use a local sbuf with SBUF_FIXEDLEN set so that mallocs do not occur while the xpt topology lock is held, regardless of the allocation policy of the passed in sbuf. Add a new routine, cam_periph_release_locked_buses(), that can be called when the caller already holds the CAM topology lock. Add some extra debugging for duplicate peripheral allocations in cam_periph_alloc(). Treat CAM_DEV_NOT_THERE much the same as a selection timeout (AC_LOST_DEVICE is emitted), but forgo retries. cam_xpt.c: Revamp the way the EDT traversal code does locking and reference counting. This was broken, since it assumed that the EDT would not change during traversal, but that assumption is no longer valid. So, to prevent devices from going away while we traverse the EDT, make sure we properly lock everything and hold references on devices that we are using. The two peripheral driver traversal routines should be examined. xptpdperiphtraverse() holds the topology lock for the entire time it runs. xptperiphtraverse() is now locked properly, but only holds the topology lock while it is traversing the list, and not while the traversal function is running. The bus locking code in xptbustraverse() should also be revisited at a later time, since it is complex and should probably be simplified. scsi_da.c: Pay attention to the return value from cam_periph_acquire(). Return 0 always from daclose() even if the disk is now gone. Add some rudimentary error injection support. scsi_sg.c: Fix reference counting in the sg(4) driver. The sg driver was calling cam_periph_release() on close, but never called cam_periph_acquire() (which increments the reference count) on open. The periph code correctly complained that the sg(4) driver was trying to decrement the refcount when it was already 0. Sponsored by: Spectra Logic MFC after: 2 weeks
2012-01-12 00:41:48 +00:00
/* increase the refcount */
if (cam_periph_acquire(periph) == CAM_REQ_CMP) {
taskqueue_enqueue(taskqueue_thread,
&softc->sysctl_task);
xpt_announce_periph(periph, announce_buf);
Added output of device QUIRKS for CAM and AHCI devices during boot. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-05-18 23:36:21 +00:00
xpt_announce_quirks(periph, softc->quirks,
DA_Q_BIT_STRING);
Fix a race condition in CAM peripheral free handling, locking in the CAM XPT bus traversal code, and a number of other periph level issues. cam_periph.h, cam_periph.c: Modify cam_periph_acquire() to test the CAM_PERIPH_INVALID flag prior to allowing a reference count to be gained on a peripheral. Callers of this function will receive CAM_REQ_CMP_ERR status in the situation of attempting to reference an invalidated periph. This guarantees that a peripheral scheduled for a deferred free will not be accessed during its wait for destruction. Panic during attempts to drop a reference count on a peripheral that already has a zero reference count. In cam_periph_list(), use a local sbuf with SBUF_FIXEDLEN set so that mallocs do not occur while the xpt topology lock is held, regardless of the allocation policy of the passed in sbuf. Add a new routine, cam_periph_release_locked_buses(), that can be called when the caller already holds the CAM topology lock. Add some extra debugging for duplicate peripheral allocations in cam_periph_alloc(). Treat CAM_DEV_NOT_THERE much the same as a selection timeout (AC_LOST_DEVICE is emitted), but forgo retries. cam_xpt.c: Revamp the way the EDT traversal code does locking and reference counting. This was broken, since it assumed that the EDT would not change during traversal, but that assumption is no longer valid. So, to prevent devices from going away while we traverse the EDT, make sure we properly lock everything and hold references on devices that we are using. The two peripheral driver traversal routines should be examined. xptpdperiphtraverse() holds the topology lock for the entire time it runs. xptperiphtraverse() is now locked properly, but only holds the topology lock while it is traversing the list, and not while the traversal function is running. The bus locking code in xptbustraverse() should also be revisited at a later time, since it is complex and should probably be simplified. scsi_da.c: Pay attention to the return value from cam_periph_acquire(). Return 0 always from daclose() even if the disk is now gone. Add some rudimentary error injection support. scsi_sg.c: Fix reference counting in the sg(4) driver. The sg driver was calling cam_periph_release() on close, but never called cam_periph_acquire() (which increments the reference count) on open. The periph code correctly complained that the sg(4) driver was trying to decrement the refcount when it was already 0. Sponsored by: Spectra Logic MFC after: 2 weeks
2012-01-12 00:41:48 +00:00
} else {
xpt_print(periph->path, "fatal error, "
"could not acquire reference count\n");
}
Move dynamic sysctl(8) variable creation for the cd(4) and da(4) drivers out of cdregister() and daregister(), which are run from interrupt context. The sysctl code does blocking mallocs (M_WAITOK), which causes problems if malloc(9) actually needs to sleep. The eventual fix for this issue will involve moving the CAM probe process inside a kernel thread. For now, though, I have fixed the issue by moving dynamic sysctl variable creation for these two drivers to a task queue running in a kernel thread. The existing task queues (taskqueue_swi and taskqueue_swi_giant) run in software interrupt handlers, which wouldn't fix the problem at hand. So I have created a new task queue, taskqueue_thread, that runs inside a kernel thread. (It also runs outside of Giant -- clients must explicitly acquire and release Giant in their taskqueue functions.) scsi_cd.c: Remove sysctl variable creation code from cdregister(), and move it to a new function, cdsysctlinit(). Queue cdsysctlinit() to the taskqueue_thread taskqueue once we have fully registered the cd(4) driver instance. scsi_da.c: Remove sysctl variable creation code from daregister(), and move it to move it to a new function, dasysctlinit(). Queue dasysctlinit() to the taskqueue_thread taskqueue once we have fully registered the da(4) instance. taskqueue.h: Declare the new taskqueue_thread taskqueue, update some comments. subr_taskqueue.c: Create the new kernel thread taskqueue. This taskqueue runs outside of Giant, so any functions queued to it would need to explicitly acquire/release Giant if they need it. cd.4: Update the cd(4) man page to talk about the minimum command size sysctl/loader tunable. Also note that the changer variables are available as loader tunables as well. da.4: Update the da(4) man page to cover the retry_count, default_timeout and minimum_cmd_size sysctl variables/loader tunables. Remove references to /dev/r???, they aren't used any longer. cd.9: Update the cd(9) man page to describe the CD_Q_10_BYTE_ONLY quirk. taskqueue.9: Update the taskqueue(9) man page to describe the new thread task queue, and the taskqueue_swi_giant queue. MFC after: 3 days
2003-09-03 04:46:28 +00:00
}
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
Do not reread SCSI disk VPD pages on every device open. Instead of rereading VPD pages on every device open, do it only on initial device probe, and in cases when device reported via UNIT ATTENTIONs that something has changed. Capacity is still rereaded on every open because it is more critical for operation and more probable to change in run time. On my tests with Intel 530 SSDs on mps(4) HBA this change reduces time GEOM needs to retaste the device (that includes few open/close cycles) from ~150ms to ~30ms. MFC after: 2 weeks Sponsored by: iXsystems, Inc.
2014-04-30 17:38:26 +00:00
/* We already probed the device. */
if (softc->flags & DA_FLAG_PROBED) {
daprobedone(periph, done_ccb);
return;
}
Update probe flow so that devices with lbp can also disable disksort. Ensure that delete_available is reset so re-probes after a media change, to one with different delete characteristics, will result in the correct methods being flagged as available. Make all ccb state changes use a consistent flow: * free() * xpt_release_ccb() * softc->state = <new state> * xpt_schedule() Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 14:14:57 +00:00
/* Ensure re-probe doesn't see old delete. */
softc->delete_available = 0;
Make delete method set via kern.cam.da.X.delete_method persistent. This allows to set delete method via tunable, before device capabilities are known. Also allow ZERO method for devices not reporting LBP, if user explicitly requests it -- it may be useful if storage supports compression and WRITE SAME, but does not support UNMAP. MFC after: 2 weeks
2015-10-11 18:26:06 +00:00
dadeleteflag(softc, DA_DELETE_ZERO, 1);
Disable UNMAP support for STEC 842 SSDs. In some unknown cases UNMAP commands make device firmware stuck. MFC after: 2 weeks
2014-04-23 19:50:35 +00:00
if (lbp && (softc->quirks & DA_Q_NO_UNMAP) == 0) {
Update probe flow so that devices with lbp can also disable disksort. Ensure that delete_available is reset so re-probes after a media change, to one with different delete characteristics, will result in the correct methods being flagged as available. Make all ccb state changes use a consistent flow: * free() * xpt_release_ccb() * softc->state = <new state> * xpt_schedule() Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 14:14:57 +00:00
/*
* Based on older SBC-3 spec revisions
* any of the UNMAP methods "may" be
* available via LBP given this flag so
* we flag all of them as availble and
* then remove those which further
* probes confirm aren't available
* later.
*
* We could also check readcap(16) p_type
* flag to exclude one or more invalid
* write same (X) types here
*/
dadeleteflag(softc, DA_DELETE_WS16, 1);
dadeleteflag(softc, DA_DELETE_WS10, 1);
dadeleteflag(softc, DA_DELETE_UNMAP, 1);
xpt_release_ccb(done_ccb);
softc->state = DA_STATE_PROBE_LBP;
xpt_schedule(periph, priority);
return;
}
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
xpt_release_ccb(done_ccb);
Added automatic detection of non-rotating media which disables the use of BIO queue sorting, hence optimising performance for devices such as SSD's Reviewed by: scottl Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:31:03 +00:00
softc->state = DA_STATE_PROBE_BDC;
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
xpt_schedule(periph, priority);
return;
}
case DA_CCB_PROBE_LBP:
{
struct scsi_vpd_logical_block_prov *lbp;
lbp = (struct scsi_vpd_logical_block_prov *)csio->data_ptr;
if ((csio->ccb_h.status & CAM_STATUS_MASK) == CAM_REQ_CMP) {
/*
* T10/1799-D Revision 31 states at least one of these
* must be supported but we don't currently enforce this.
*/
dadeleteflag(softc, DA_DELETE_WS16,
(lbp->flags & SVPD_LBP_WS16));
dadeleteflag(softc, DA_DELETE_WS10,
(lbp->flags & SVPD_LBP_WS10));
dadeleteflag(softc, DA_DELETE_UNMAP,
(lbp->flags & SVPD_LBP_UNMAP));
} else {
int error;
error = daerror(done_ccb, CAM_RETRY_SELTO,
SF_RETRY_UA|SF_NO_PRINT);
if (error == ERESTART)
return;
else if (error != 0) {
if ((done_ccb->ccb_h.status & CAM_DEV_QFRZN) != 0) {
/* Don't wedge this device's queue */
cam_release_devq(done_ccb->ccb_h.path,
/*relsim_flags*/0,
/*reduction*/0,
/*timeout*/0,
/*getcount_only*/0);
}
/*
* Failure indicates we don't support any SBC-3
* delete methods with UNMAP
*/
}
}
free(lbp, M_SCSIDA);
xpt_release_ccb(done_ccb);
Respect MAXIMUM TRANSFER LENGTH field of Block Limits VPD page. Nobody yet reported disk supporting I/Os less then our MAXPHYS value, but since we any way have code to read Block Limits VPD page, that is easy. MFC after: 2 weeks
2014-04-30 19:44:31 +00:00
softc->state = DA_STATE_PROBE_BLK_LIMITS;
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
xpt_schedule(periph, priority);
return;
}
case DA_CCB_PROBE_BLK_LIMITS:
{
struct scsi_vpd_block_limits *block_limits;
block_limits = (struct scsi_vpd_block_limits *)csio->data_ptr;
if ((csio->ccb_h.status & CAM_STATUS_MASK) == CAM_REQ_CMP) {
Respect MAXIMUM TRANSFER LENGTH field of Block Limits VPD page. Nobody yet reported disk supporting I/Os less then our MAXPHYS value, but since we any way have code to read Block Limits VPD page, that is easy. MFC after: 2 weeks
2014-04-30 19:44:31 +00:00
uint32_t max_txfer_len = scsi_4btoul(
block_limits->max_txfer_len);
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
uint32_t max_unmap_lba_cnt = scsi_4btoul(
block_limits->max_unmap_lba_cnt);
uint32_t max_unmap_blk_cnt = scsi_4btoul(
block_limits->max_unmap_blk_cnt);
uint64_t ws_max_blks = scsi_8btou64(
block_limits->max_write_same_length);
Respect MAXIMUM TRANSFER LENGTH field of Block Limits VPD page. Nobody yet reported disk supporting I/Os less then our MAXPHYS value, but since we any way have code to read Block Limits VPD page, that is easy. MFC after: 2 weeks
2014-04-30 19:44:31 +00:00
if (max_txfer_len != 0) {
softc->disk->d_maxsize = MIN(softc->maxio,
(off_t)max_txfer_len * softc->params.secsize);
}
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
/*
* We should already support UNMAP but we check lba
* and block count to be sure
*/
if (max_unmap_lba_cnt != 0x00L &&
max_unmap_blk_cnt != 0x00L) {
softc->unmap_max_lba = max_unmap_lba_cnt;
softc->unmap_max_ranges = min(max_unmap_blk_cnt,
UNMAP_MAX_RANGES);
} else {
/*
* Unexpected UNMAP limits which means the
* device doesn't actually support UNMAP
*/
dadeleteflag(softc, DA_DELETE_UNMAP, 0);
}
if (ws_max_blks != 0x00L)
softc->ws_max_blks = ws_max_blks;
} else {
int error;
error = daerror(done_ccb, CAM_RETRY_SELTO,
SF_RETRY_UA|SF_NO_PRINT);
if (error == ERESTART)
return;
else if (error != 0) {
if ((done_ccb->ccb_h.status & CAM_DEV_QFRZN) != 0) {
/* Don't wedge this device's queue */
cam_release_devq(done_ccb->ccb_h.path,
/*relsim_flags*/0,
/*reduction*/0,
/*timeout*/0,
/*getcount_only*/0);
}
/*
* Failure here doesn't mean UNMAP is not
* supported as this is an optional page.
*/
softc->unmap_max_lba = 1;
softc->unmap_max_ranges = 1;
}
}
free(block_limits, M_SCSIDA);
xpt_release_ccb(done_ccb);
Update probe flow so that devices with lbp can also disable disksort. Ensure that delete_available is reset so re-probes after a media change, to one with different delete characteristics, will result in the correct methods being flagged as available. Make all ccb state changes use a consistent flow: * free() * xpt_release_ccb() * softc->state = <new state> * xpt_schedule() Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 14:14:57 +00:00
softc->state = DA_STATE_PROBE_BDC;
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
xpt_schedule(periph, priority);
return;
}
Added automatic detection of non-rotating media which disables the use of BIO queue sorting, hence optimising performance for devices such as SSD's Reviewed by: scottl Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:31:03 +00:00
case DA_CCB_PROBE_BDC:
{
struct scsi_vpd_block_characteristics *bdc;
bdc = (struct scsi_vpd_block_characteristics *)csio->data_ptr;
if ((csio->ccb_h.status & CAM_STATUS_MASK) == CAM_REQ_CMP) {
Correct comment typo's Add missing comment Reviewed by: pjd (mentor) Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-28 21:14:23 +00:00
/*
* Disable queue sorting for non-rotational media
* by default.
*/
Corrected definition for old_rate to match d_rotation_rate MFC after: 2 Days X-MFC-With: r256956
2013-11-07 23:21:52 +00:00
u_int16_t old_rate = softc->disk->d_rotation_rate;
Improve ZFS N-way mirror read performance by using load and locality information. The existing algorithm selects a preferred leaf vdev based on offset of the zio request modulo the number of members in the mirror. It assumes the devices are of equal performance and that spreading the requests randomly over both drives will be sufficient to saturate them. In practice this results in the leaf vdevs being under utilized. The new algorithm takes into the following additional factors: * Load of the vdevs (number outstanding I/O requests) * The locality of last queued I/O vs the new I/O request. Within the locality calculation additional knowledge about the underlying vdev is considered such as; is the device backing the vdev a rotating media device. This results in performance increases across the board as well as significant increases for predominantly streaming loads and for configurations which don't have evenly performing devices. The following are results from a setup with 3 Way Mirror with 2 x HD's and 1 x SSD from a basic test running multiple parrallel dd's. With pre-fetch disabled (vfs.zfs.prefetch_disable=1): == Stripe Balanced (default) == Read 15360MB using bs: 1048576, readers: 3, took 161 seconds @ 95 MB/s == Load Balanced (zfslinux) == Read 15360MB using bs: 1048576, readers: 3, took 297 seconds @ 51 MB/s == Load Balanced (locality freebsd) == Read 15360MB using bs: 1048576, readers: 3, took 54 seconds @ 284 MB/s With pre-fetch enabled (vfs.zfs.prefetch_disable=0): == Stripe Balanced (default) == Read 15360MB using bs: 1048576, readers: 3, took 91 seconds @ 168 MB/s == Load Balanced (zfslinux) == Read 15360MB using bs: 1048576, readers: 3, took 108 seconds @ 142 MB/s == Load Balanced (locality freebsd) == Read 15360MB using bs: 1048576, readers: 3, took 48 seconds @ 320 MB/s In addition to the performance changes the code was also restructured, with the help of Justin Gibbs, to provide a more logical flow which also ensures vdevs loads are only calculated from the set of valid candidates. The following additional sysctls where added to allow the administrator to tune the behaviour of the load algorithm: * vfs.zfs.vdev.mirror.rotating_inc * vfs.zfs.vdev.mirror.rotating_seek_inc * vfs.zfs.vdev.mirror.rotating_seek_offset * vfs.zfs.vdev.mirror.non_rotating_inc * vfs.zfs.vdev.mirror.non_rotating_seek_inc These changes where based on work started by the zfsonlinux developers: https://github.com/zfsonlinux/zfs/pull/1487 Reviewed by: gibbs, mav, will MFC after: 2 weeks Sponsored by: Multiplay
2013-10-23 09:54:58 +00:00
softc->disk->d_rotation_rate =
scsi_2btoul(bdc->medium_rotation_rate);
if (softc->disk->d_rotation_rate ==
SVPD_BDC_RATE_NON_ROTATING) {
Added automatic detection of non-rotating media which disables the use of BIO queue sorting, hence optimising performance for devices such as SSD's Reviewed by: scottl Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:31:03 +00:00
softc->sort_io_queue = 0;
Improve ZFS N-way mirror read performance by using load and locality information. The existing algorithm selects a preferred leaf vdev based on offset of the zio request modulo the number of members in the mirror. It assumes the devices are of equal performance and that spreading the requests randomly over both drives will be sufficient to saturate them. In practice this results in the leaf vdevs being under utilized. The new algorithm takes into the following additional factors: * Load of the vdevs (number outstanding I/O requests) * The locality of last queued I/O vs the new I/O request. Within the locality calculation additional knowledge about the underlying vdev is considered such as; is the device backing the vdev a rotating media device. This results in performance increases across the board as well as significant increases for predominantly streaming loads and for configurations which don't have evenly performing devices. The following are results from a setup with 3 Way Mirror with 2 x HD's and 1 x SSD from a basic test running multiple parrallel dd's. With pre-fetch disabled (vfs.zfs.prefetch_disable=1): == Stripe Balanced (default) == Read 15360MB using bs: 1048576, readers: 3, took 161 seconds @ 95 MB/s == Load Balanced (zfslinux) == Read 15360MB using bs: 1048576, readers: 3, took 297 seconds @ 51 MB/s == Load Balanced (locality freebsd) == Read 15360MB using bs: 1048576, readers: 3, took 54 seconds @ 284 MB/s With pre-fetch enabled (vfs.zfs.prefetch_disable=0): == Stripe Balanced (default) == Read 15360MB using bs: 1048576, readers: 3, took 91 seconds @ 168 MB/s == Load Balanced (zfslinux) == Read 15360MB using bs: 1048576, readers: 3, took 108 seconds @ 142 MB/s == Load Balanced (locality freebsd) == Read 15360MB using bs: 1048576, readers: 3, took 48 seconds @ 320 MB/s In addition to the performance changes the code was also restructured, with the help of Justin Gibbs, to provide a more logical flow which also ensures vdevs loads are only calculated from the set of valid candidates. The following additional sysctls where added to allow the administrator to tune the behaviour of the load algorithm: * vfs.zfs.vdev.mirror.rotating_inc * vfs.zfs.vdev.mirror.rotating_seek_inc * vfs.zfs.vdev.mirror.rotating_seek_offset * vfs.zfs.vdev.mirror.non_rotating_inc * vfs.zfs.vdev.mirror.non_rotating_seek_inc These changes where based on work started by the zfsonlinux developers: https://github.com/zfsonlinux/zfs/pull/1487 Reviewed by: gibbs, mav, will MFC after: 2 weeks Sponsored by: Multiplay
2013-10-23 09:54:58 +00:00
}
if (softc->disk->d_rotation_rate != old_rate) {
disk_attr_changed(softc->disk,
"GEOM::rotation_rate", M_NOWAIT);
}
Added automatic detection of non-rotating media which disables the use of BIO queue sorting, hence optimising performance for devices such as SSD's Reviewed by: scottl Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:31:03 +00:00
} else {
int error;
error = daerror(done_ccb, CAM_RETRY_SELTO,
SF_RETRY_UA|SF_NO_PRINT);
if (error == ERESTART)
return;
else if (error != 0) {
if ((done_ccb->ccb_h.status & CAM_DEV_QFRZN) != 0) {
/* Don't wedge this device's queue */
cam_release_devq(done_ccb->ccb_h.path,
/*relsim_flags*/0,
/*reduction*/0,
/*timeout*/0,
/*getcount_only*/0);
}
}
}
free(bdc, M_SCSIDA);
xpt_release_ccb(done_ccb);
Update probe flow so that devices with lbp can also disable disksort. Ensure that delete_available is reset so re-probes after a media change, to one with different delete characteristics, will result in the correct methods being flagged as available. Make all ccb state changes use a consistent flow: * free() * xpt_release_ccb() * softc->state = <new state> * xpt_schedule() Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 14:14:57 +00:00
softc->state = DA_STATE_PROBE_ATA;
Added automatic detection of non-rotating media which disables the use of BIO queue sorting, hence optimising performance for devices such as SSD's Reviewed by: scottl Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:31:03 +00:00
xpt_schedule(periph, priority);
return;
}
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
case DA_CCB_PROBE_ATA:
{
int i;
struct ata_params *ata_params;
int16_t *ptr;
ata_params = (struct ata_params *)csio->data_ptr;
ptr = (uint16_t *)ata_params;
if ((csio->ccb_h.status & CAM_STATUS_MASK) == CAM_REQ_CMP) {
Improve ZFS N-way mirror read performance by using load and locality information. The existing algorithm selects a preferred leaf vdev based on offset of the zio request modulo the number of members in the mirror. It assumes the devices are of equal performance and that spreading the requests randomly over both drives will be sufficient to saturate them. In practice this results in the leaf vdevs being under utilized. The new algorithm takes into the following additional factors: * Load of the vdevs (number outstanding I/O requests) * The locality of last queued I/O vs the new I/O request. Within the locality calculation additional knowledge about the underlying vdev is considered such as; is the device backing the vdev a rotating media device. This results in performance increases across the board as well as significant increases for predominantly streaming loads and for configurations which don't have evenly performing devices. The following are results from a setup with 3 Way Mirror with 2 x HD's and 1 x SSD from a basic test running multiple parrallel dd's. With pre-fetch disabled (vfs.zfs.prefetch_disable=1): == Stripe Balanced (default) == Read 15360MB using bs: 1048576, readers: 3, took 161 seconds @ 95 MB/s == Load Balanced (zfslinux) == Read 15360MB using bs: 1048576, readers: 3, took 297 seconds @ 51 MB/s == Load Balanced (locality freebsd) == Read 15360MB using bs: 1048576, readers: 3, took 54 seconds @ 284 MB/s With pre-fetch enabled (vfs.zfs.prefetch_disable=0): == Stripe Balanced (default) == Read 15360MB using bs: 1048576, readers: 3, took 91 seconds @ 168 MB/s == Load Balanced (zfslinux) == Read 15360MB using bs: 1048576, readers: 3, took 108 seconds @ 142 MB/s == Load Balanced (locality freebsd) == Read 15360MB using bs: 1048576, readers: 3, took 48 seconds @ 320 MB/s In addition to the performance changes the code was also restructured, with the help of Justin Gibbs, to provide a more logical flow which also ensures vdevs loads are only calculated from the set of valid candidates. The following additional sysctls where added to allow the administrator to tune the behaviour of the load algorithm: * vfs.zfs.vdev.mirror.rotating_inc * vfs.zfs.vdev.mirror.rotating_seek_inc * vfs.zfs.vdev.mirror.rotating_seek_offset * vfs.zfs.vdev.mirror.non_rotating_inc * vfs.zfs.vdev.mirror.non_rotating_seek_inc These changes where based on work started by the zfsonlinux developers: https://github.com/zfsonlinux/zfs/pull/1487 Reviewed by: gibbs, mav, will MFC after: 2 weeks Sponsored by: Multiplay
2013-10-23 09:54:58 +00:00
uint16_t old_rate;
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
for (i = 0; i < sizeof(*ata_params) / 2; i++)
ptr[i] = le16toh(ptr[i]);
Disable UNMAP support for STEC 842 SSDs. In some unknown cases UNMAP commands make device firmware stuck. MFC after: 2 weeks
2014-04-23 19:50:35 +00:00
if (ata_params->support_dsm & ATA_SUPPORT_DSM_TRIM &&
(softc->quirks & DA_Q_NO_UNMAP) == 0) {
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
dadeleteflag(softc, DA_DELETE_ATA_TRIM, 1);
if (ata_params->max_dsm_blocks != 0)
softc->trim_max_ranges = min(
softc->trim_max_ranges,
ata_params->max_dsm_blocks *
ATA_DSM_BLK_RANGES);
}
Added automatic detection of non-rotating media which disables the use of BIO queue sorting, hence optimising performance for devices such as SSD's Reviewed by: scottl Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:31:03 +00:00
/*
Correct comment typo's Add missing comment Reviewed by: pjd (mentor) Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-28 21:14:23 +00:00
* Disable queue sorting for non-rotational media
* by default.
Added automatic detection of non-rotating media which disables the use of BIO queue sorting, hence optimising performance for devices such as SSD's Reviewed by: scottl Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:31:03 +00:00
*/
Improve ZFS N-way mirror read performance by using load and locality information. The existing algorithm selects a preferred leaf vdev based on offset of the zio request modulo the number of members in the mirror. It assumes the devices are of equal performance and that spreading the requests randomly over both drives will be sufficient to saturate them. In practice this results in the leaf vdevs being under utilized. The new algorithm takes into the following additional factors: * Load of the vdevs (number outstanding I/O requests) * The locality of last queued I/O vs the new I/O request. Within the locality calculation additional knowledge about the underlying vdev is considered such as; is the device backing the vdev a rotating media device. This results in performance increases across the board as well as significant increases for predominantly streaming loads and for configurations which don't have evenly performing devices. The following are results from a setup with 3 Way Mirror with 2 x HD's and 1 x SSD from a basic test running multiple parrallel dd's. With pre-fetch disabled (vfs.zfs.prefetch_disable=1): == Stripe Balanced (default) == Read 15360MB using bs: 1048576, readers: 3, took 161 seconds @ 95 MB/s == Load Balanced (zfslinux) == Read 15360MB using bs: 1048576, readers: 3, took 297 seconds @ 51 MB/s == Load Balanced (locality freebsd) == Read 15360MB using bs: 1048576, readers: 3, took 54 seconds @ 284 MB/s With pre-fetch enabled (vfs.zfs.prefetch_disable=0): == Stripe Balanced (default) == Read 15360MB using bs: 1048576, readers: 3, took 91 seconds @ 168 MB/s == Load Balanced (zfslinux) == Read 15360MB using bs: 1048576, readers: 3, took 108 seconds @ 142 MB/s == Load Balanced (locality freebsd) == Read 15360MB using bs: 1048576, readers: 3, took 48 seconds @ 320 MB/s In addition to the performance changes the code was also restructured, with the help of Justin Gibbs, to provide a more logical flow which also ensures vdevs loads are only calculated from the set of valid candidates. The following additional sysctls where added to allow the administrator to tune the behaviour of the load algorithm: * vfs.zfs.vdev.mirror.rotating_inc * vfs.zfs.vdev.mirror.rotating_seek_inc * vfs.zfs.vdev.mirror.rotating_seek_offset * vfs.zfs.vdev.mirror.non_rotating_inc * vfs.zfs.vdev.mirror.non_rotating_seek_inc These changes where based on work started by the zfsonlinux developers: https://github.com/zfsonlinux/zfs/pull/1487 Reviewed by: gibbs, mav, will MFC after: 2 weeks Sponsored by: Multiplay
2013-10-23 09:54:58 +00:00
old_rate = softc->disk->d_rotation_rate;
softc->disk->d_rotation_rate =
ata_params->media_rotation_rate;
if (softc->disk->d_rotation_rate ==
ATA_RATE_NON_ROTATING) {
Added automatic detection of non-rotating media which disables the use of BIO queue sorting, hence optimising performance for devices such as SSD's Reviewed by: scottl Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:31:03 +00:00
softc->sort_io_queue = 0;
Improve ZFS N-way mirror read performance by using load and locality information. The existing algorithm selects a preferred leaf vdev based on offset of the zio request modulo the number of members in the mirror. It assumes the devices are of equal performance and that spreading the requests randomly over both drives will be sufficient to saturate them. In practice this results in the leaf vdevs being under utilized. The new algorithm takes into the following additional factors: * Load of the vdevs (number outstanding I/O requests) * The locality of last queued I/O vs the new I/O request. Within the locality calculation additional knowledge about the underlying vdev is considered such as; is the device backing the vdev a rotating media device. This results in performance increases across the board as well as significant increases for predominantly streaming loads and for configurations which don't have evenly performing devices. The following are results from a setup with 3 Way Mirror with 2 x HD's and 1 x SSD from a basic test running multiple parrallel dd's. With pre-fetch disabled (vfs.zfs.prefetch_disable=1): == Stripe Balanced (default) == Read 15360MB using bs: 1048576, readers: 3, took 161 seconds @ 95 MB/s == Load Balanced (zfslinux) == Read 15360MB using bs: 1048576, readers: 3, took 297 seconds @ 51 MB/s == Load Balanced (locality freebsd) == Read 15360MB using bs: 1048576, readers: 3, took 54 seconds @ 284 MB/s With pre-fetch enabled (vfs.zfs.prefetch_disable=0): == Stripe Balanced (default) == Read 15360MB using bs: 1048576, readers: 3, took 91 seconds @ 168 MB/s == Load Balanced (zfslinux) == Read 15360MB using bs: 1048576, readers: 3, took 108 seconds @ 142 MB/s == Load Balanced (locality freebsd) == Read 15360MB using bs: 1048576, readers: 3, took 48 seconds @ 320 MB/s In addition to the performance changes the code was also restructured, with the help of Justin Gibbs, to provide a more logical flow which also ensures vdevs loads are only calculated from the set of valid candidates. The following additional sysctls where added to allow the administrator to tune the behaviour of the load algorithm: * vfs.zfs.vdev.mirror.rotating_inc * vfs.zfs.vdev.mirror.rotating_seek_inc * vfs.zfs.vdev.mirror.rotating_seek_offset * vfs.zfs.vdev.mirror.non_rotating_inc * vfs.zfs.vdev.mirror.non_rotating_seek_inc These changes where based on work started by the zfsonlinux developers: https://github.com/zfsonlinux/zfs/pull/1487 Reviewed by: gibbs, mav, will MFC after: 2 weeks Sponsored by: Multiplay
2013-10-23 09:54:58 +00:00
}
if (softc->disk->d_rotation_rate != old_rate) {
disk_attr_changed(softc->disk,
"GEOM::rotation_rate", M_NOWAIT);
}
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
} else {
int error;
error = daerror(done_ccb, CAM_RETRY_SELTO,
Use the existence of ATA Information VPD to determine if we should attempt to query ATA functionality via ATA Pass-Through (16) as this page is defined as "must" for SATL devices, hence indicating that the device is at least likely to support Pass-Through (16). This eliminates errors produced by CTL when ATA Pass-Through (16) fails. Switch ATA probe daerror call to SF_NO_PRINT to avoid errors printing out for devices which return invalid errors. Output details about supported and choosen delete method when verbose booted. Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 14:37:23 +00:00
SF_RETRY_UA|SF_NO_PRINT);
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
if (error == ERESTART)
return;
else if (error != 0) {
if ((done_ccb->ccb_h.status & CAM_DEV_QFRZN) != 0) {
/* Don't wedge this device's queue */
cam_release_devq(done_ccb->ccb_h.path,
/*relsim_flags*/0,
/*reduction*/0,
/*timeout*/0,
/*getcount_only*/0);
}
}
}
free(ata_params, M_SCSIDA);
Use the existence of ATA Information VPD to determine if we should attempt to query ATA functionality via ATA Pass-Through (16) as this page is defined as "must" for SATL devices, hence indicating that the device is at least likely to support Pass-Through (16). This eliminates errors produced by CTL when ATA Pass-Through (16) fails. Switch ATA probe daerror call to SF_NO_PRINT to avoid errors printing out for devices which return invalid errors. Output details about supported and choosen delete method when verbose booted. Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 14:37:23 +00:00
daprobedone(periph, done_ccb);
A better fix to avoid race conditions between failed probes and peripheral removal. Obtained from:gibbs@freebsd.org
1999-01-07 20:19:09 +00:00
return;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
case DA_CCB_DUMP:
/* No-op. We're polling */
return;
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
case DA_CCB_TUR:
{
if ((done_ccb->ccb_h.status & CAM_STATUS_MASK) != CAM_REQ_CMP) {
if (daerror(done_ccb, CAM_RETRY_SELTO,
SF_RETRY_UA | SF_NO_RECOVERY | SF_NO_PRINT) ==
ERESTART)
return;
if ((done_ccb->ccb_h.status & CAM_DEV_QFRZN) != 0)
cam_release_devq(done_ccb->ccb_h.path,
/*relsim_flags*/0,
/*reduction*/0,
/*timeout*/0,
/*getcount_only*/0);
}
xpt_release_ccb(done_ccb);
cam_periph_release_locked(periph);
return;
}
A better fix to avoid race conditions between failed probes and peripheral removal. Obtained from:gibbs@freebsd.org
1999-01-07 20:19:09 +00:00
default:
break;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
A better fix to avoid race conditions between failed probes and peripheral removal. Obtained from:gibbs@freebsd.org
1999-01-07 20:19:09 +00:00
xpt_release_ccb(done_ccb);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
Refactor da(4) to remove one of two code paths used to query capacity data. Reviewed by: ken, mav (earlier version) Sponsored by: The FreeBSD Foundation
2012-04-12 12:58:14 +00:00
static void
dareprobe(struct cam_periph *periph)
{
struct da_softc *softc;
cam_status status;
softc = (struct da_softc *)periph->softc;
/* Probe in progress; don't interfere. */
Fix probe in progress check in dareprobe Reviewed by: mav Approved by: pjd (mentor) MFC after: 1 week
2013-05-02 14:19:42 +00:00
if (softc->state != DA_STATE_NORMAL)
Refactor da(4) to remove one of two code paths used to query capacity data. Reviewed by: ken, mav (earlier version) Sponsored by: The FreeBSD Foundation
2012-04-12 12:58:14 +00:00
return;
status = cam_periph_acquire(periph);
KASSERT(status == CAM_REQ_CMP,
("dareprobe: cam_periph_acquire failed"));
if (softc->flags & DA_FLAG_CAN_RC16)
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
softc->state = DA_STATE_PROBE_RC16;
Refactor da(4) to remove one of two code paths used to query capacity data. Reviewed by: ken, mav (earlier version) Sponsored by: The FreeBSD Foundation
2012-04-12 12:58:14 +00:00
else
Added available delete methods discovery during device probe, including the maximum sizes for said methods, which are used when processing BIO_DELETE requests. This includes updating UNMAP support discovery to be based on SBC-3 T10/1799-D Revision 31 specification. Added ATA TRIM support to cam scsi devices via ATA Pass-Through(16) sys/cam/scsi/scsi_da.c: - Added ATA Data Set Management TRIM support via ATA Pass-Through(16) as a delete_method - Added four new probe states used to identity available methods and their limits for the processing of BIO_DELETE commands via both UNMAP and the new ATA TRIM commands. - Renamed Probe states to better indicate their use - Added delete method descriptions used when informing user of issues. - Added automatic calculation of the optimum delete mode based on which method presents the largest maximum request size as this is most likely to result in the best performance. - Added WRITE SAME max block limits - Updated UNMAP range generation to mirror that used by ATA TRIM, this optimises the generation of ranges and fixes a potential overflow issue in the count when combining multiple BIO_DELETE requests - Added output of warnings about short deletes. This should only ever be triggered on devices that fail to correctly advertise their supported delete modes / max sizes. - Fixed WS16 requests being incorrectly limited to 65535 in length. Reviewed by: mav Approved by: pjd (mentor) MFC after: 2 weeks
2013-04-26 16:17:04 +00:00
softc->state = DA_STATE_PROBE_RC;
Refactor da(4) to remove one of two code paths used to query capacity data. Reviewed by: ken, mav (earlier version) Sponsored by: The FreeBSD Foundation
2012-04-12 12:58:14 +00:00
xpt_schedule(periph, CAM_PRIORITY_DEV);
}
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
static int
daerror(union ccb *ccb, u_int32_t cam_flags, u_int32_t sense_flags)
{
struct da_softc *softc;
struct cam_periph *periph;
Add scsi_extract_sense_ccb() -- wrapper around scsi_extract_sense_len(). It allows to remove number of duplicate checks from several places.
2012-06-23 12:32:53 +00:00
int error, error_code, sense_key, asc, ascq;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
periph = xpt_path_periph(ccb->ccb_h.path);
softc = (struct da_softc *)periph->softc;
Automatically detect devices that do not support READ(6)/WRITE(6) and upgrade to using 10 byte cdbs. As far as I tested, this works efficiently for most of the SBP-II/Firewire devices but most of the umass devices still need ad-hoc work around because umass-sim doesn't return any SCSI errors. A sysctl nob is also added for the last resort. I hope we don't need DA_Q_NO_6_BYTE quirks anymore. Reviewed by: gibbs MFC after: 1 week
2002-03-23 18:18:02 +00:00
/*
* Automatically detect devices that do not support
* READ(6)/WRITE(6) and upgrade to using 10 byte cdbs.
*/
error = 0;
Fix ATAPI/USB/Firewire CDROM drive handling in cd(4) and hopefully fix a number of related problems along the way. - Automatically detect CDROM drives that can't handle 6 byte mode sense and mode select, and adjust our command size accordingly. We have to handle this in the cd(4) driver (where the buffers are allocated), since the parameter list length is different for the 6 and 10 byte mode sense commands. - Remove MODE_SENSE and MODE_SELECT translation removed in ATAPICAM and in the umass(4) driver, since there's no way for that to work properly. - Add a quirk entry for CDROM drives that just hang when they get a 6 byte mode sense or mode select. The reason for the quirk must be documented in a PR, and all quirks must be approved by ken@FreeBSD.org. This is to make sure that we fully understand why each quirk is needed. Once the CAM_NEW_TRAN_CODE is finished, we should be able to remove any such quirks, since we'll know what protocol the drive speaks (SCSI, ATAPI, etc.) and therefore whether we should use 6 or 10 byte mode sense/select commands. - Change the way the da(4) handles the no_6_byte sysctl. There is now a per-drive sysctl to set the minimum command size for that particular disk. (Since you could have multiple disks with multiple requirements in one system.) - Loader tunable support for all the sysctls in the da(4) and cd(4) drivers. - Add a CDIOCCLOSE ioctl for cd(4) (bde pointed this out a long time ago). - Add a media validation routine (cdcheckmedia()) to the cd(4) driver, to fix some problems bde pointed out a long time ago. We now allow open() to succeed no matter what, but if we don't detect valid media, the user can only issue CDIOCCLOSE or CDIOCEJECT ioctls. - The media validation routine also reads the table of contents off the drive. We use the table of contents to implement the CDIOCPLAYTRACKS ioctl using the PLAY AUDIO MSF command. The PLAY AUDIO TRACK INDEX command that we previously used was deprecated after SCSI-2. It works in every SCSI CDROM I've tried, but doesn't seem to work on ATAPI CDROM drives. We still use the play audio track index command if we don't have a valid TOC, but I suppose it'll fail anyway in that case. - Add _len() versions of scsi_mode_sense() and scsi_mode_select() so that we can specify the minimum command length. - Fix a couple of formatting problems in the sense printing code. MFC after: 4 weeks
2003-02-21 06:19:38 +00:00
if ((ccb->ccb_h.status & CAM_STATUS_MASK) == CAM_REQ_INVALID) {
error = cmd6workaround(ccb);
Add scsi_extract_sense_ccb() -- wrapper around scsi_extract_sense_len(). It allows to remove number of duplicate checks from several places.
2012-06-23 12:32:53 +00:00
} else if (scsi_extract_sense_ccb(ccb,
&error_code, &sense_key, &asc, &ascq)) {
Automatically detect devices that do not support READ(6)/WRITE(6) and upgrade to using 10 byte cdbs. As far as I tested, this works efficiently for most of the SBP-II/Firewire devices but most of the umass devices still need ad-hoc work around because umass-sim doesn't return any SCSI errors. A sysctl nob is also added for the last resort. I hope we don't need DA_Q_NO_6_BYTE quirks anymore. Reviewed by: gibbs MFC after: 1 week
2002-03-23 18:18:02 +00:00
if (sense_key == SSD_KEY_ILLEGAL_REQUEST)
error = cmd6workaround(ccb);
Refactor da(4) to remove one of two code paths used to query capacity data. Reviewed by: ken, mav (earlier version) Sponsored by: The FreeBSD Foundation
2012-04-12 12:58:14 +00:00
/*
* If the target replied with CAPACITY DATA HAS CHANGED UA,
* query the capacity and notify upper layers.
*/
else if (sense_key == SSD_KEY_UNIT_ATTENTION &&
asc == 0x2A && ascq == 0x09) {
Do not reread SCSI disk VPD pages on every device open. Instead of rereading VPD pages on every device open, do it only on initial device probe, and in cases when device reported via UNIT ATTENTIONs that something has changed. Capacity is still rereaded on every open because it is more critical for operation and more probable to change in run time. On my tests with Intel 530 SSDs on mps(4) HBA this change reduces time GEOM needs to retaste the device (that includes few open/close cycles) from ~150ms to ~30ms. MFC after: 2 weeks Sponsored by: iXsystems, Inc.
2014-04-30 17:38:26 +00:00
xpt_print(periph->path, "Capacity data has changed\n");
softc->flags &= ~DA_FLAG_PROBED;
Refactor da(4) to remove one of two code paths used to query capacity data. Reviewed by: ken, mav (earlier version) Sponsored by: The FreeBSD Foundation
2012-04-12 12:58:14 +00:00
dareprobe(periph);
sense_flags |= SF_NO_PRINT;
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
} else if (sense_key == SSD_KEY_UNIT_ATTENTION &&
Do not reread SCSI disk VPD pages on every device open. Instead of rereading VPD pages on every device open, do it only on initial device probe, and in cases when device reported via UNIT ATTENTIONs that something has changed. Capacity is still rereaded on every open because it is more critical for operation and more probable to change in run time. On my tests with Intel 530 SSDs on mps(4) HBA this change reduces time GEOM needs to retaste the device (that includes few open/close cycles) from ~150ms to ~30ms. MFC after: 2 weeks Sponsored by: iXsystems, Inc.
2014-04-30 17:38:26 +00:00
asc == 0x28 && ascq == 0x00) {
softc->flags &= ~DA_FLAG_PROBED;
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
disk_media_changed(softc->disk, M_NOWAIT);
Do not reread SCSI disk VPD pages on every device open. Instead of rereading VPD pages on every device open, do it only on initial device probe, and in cases when device reported via UNIT ATTENTIONs that something has changed. Capacity is still rereaded on every open because it is more critical for operation and more probable to change in run time. On my tests with Intel 530 SSDs on mps(4) HBA this change reduces time GEOM needs to retaste the device (that includes few open/close cycles) from ~150ms to ~30ms. MFC after: 2 weeks Sponsored by: iXsystems, Inc.
2014-04-30 17:38:26 +00:00
} else if (sense_key == SSD_KEY_UNIT_ATTENTION &&
asc == 0x3F && ascq == 0x03) {
xpt_print(periph->path, "INQUIRY data has changed\n");
softc->flags &= ~DA_FLAG_PROBED;
dareprobe(periph);
sense_flags |= SF_NO_PRINT;
} else if (sense_key == SSD_KEY_NOT_READY &&
Tune support for removable media in da driver: - remove DA_FLAG_SAW_MEDIA flag, almost opposite to DA_FLAG_PACK_INVALID, using the last instead. - allow opening device with no media present, reporting zero media size and non-zero sector size, as geom/notes suggests. That allow to read device attributes and potentially do other things, not related to media.
2013-05-03 11:53:06 +00:00
asc == 0x3a && (softc->flags & DA_FLAG_PACK_INVALID) == 0) {
softc->flags |= DA_FLAG_PACK_INVALID;
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
disk_media_gone(softc->disk, M_NOWAIT);
Refactor da(4) to remove one of two code paths used to query capacity data. Reviewed by: ken, mav (earlier version) Sponsored by: The FreeBSD Foundation
2012-04-12 12:58:14 +00:00
}
Automatically detect devices that do not support READ(6)/WRITE(6) and upgrade to using 10 byte cdbs. As far as I tested, this works efficiently for most of the SBP-II/Firewire devices but most of the umass devices still need ad-hoc work around because umass-sim doesn't return any SCSI errors. A sysctl nob is also added for the last resort. I hope we don't need DA_Q_NO_6_BYTE quirks anymore. Reviewed by: gibbs MFC after: 1 week
2002-03-23 18:18:02 +00:00
}
if (error == ERESTART)
Updates to cmd6workaround: * Only update cdb in place if not CDB_POINTER * Correctly check for QFRZ before restarting CCB * More accurate printf message * style(9) changes at end Patch tested successfully on Maxtor 4 G120J6 GAK8.
2002-08-16 22:05:19 +00:00
return (ERESTART);
Automatically detect devices that do not support READ(6)/WRITE(6) and upgrade to using 10 byte cdbs. As far as I tested, this works efficiently for most of the SBP-II/Firewire devices but most of the umass devices still need ad-hoc work around because umass-sim doesn't return any SCSI errors. A sysctl nob is also added for the last resort. I hope we don't need DA_Q_NO_6_BYTE quirks anymore. Reviewed by: gibbs MFC after: 1 week
2002-03-23 18:18:02 +00:00
Don't invalidate devices due to unexpected unit attention errors. In a perfect world, we'd notice the UA and do some device validation to ensure that the device hasn't changed. We may get this before the year ends, but not before 3.0R. This change gives the adminstrator ample ammunition to take off a foot or two, but hey this *is* UN*X.
1998-09-20 07:17:11 +00:00
/*
* XXX
* Until we have a better way of doing pack validation,
* don't treat UAs as errors.
*/
sense_flags |= SF_RETRY_UA;
Retry indefinitely on SCSI BUSY status from VMware disks and CDs. VMware returns BUSY status when storage has transient connectivity issues. It is often better to wait and let VM admin fix the problem then crash. Discussed with: ken MFC after: 1 week
2015-02-02 20:23:05 +00:00
if (softc->quirks & DA_Q_RETRY_BUSY)
sense_flags |= SF_RETRY_BUSY;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
return(cam_periph_error(ccb, cam_flags, sense_flags,
&softc->saved_ccb));
}
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
static void
damediapoll(void *arg)
{
struct cam_periph *periph = arg;
struct da_softc *softc = periph->softc;
Some microoptimizations for da and ada drivers: - Replace ordered_tag_count counter with single flag; - From da remove outstanding_cmds counter, duplicating pending_ccbs list; - From da_softc remove unused links field.
2013-10-24 14:05:44 +00:00
if (!softc->tur && LIST_EMPTY(&softc->pending_ccbs)) {
Implement media change notification for DA and CD removable media devices. It includes three parts: 1) Modifications to CAM to detect media media changes and report them to disk(9) layer. For modern SATA (and potentially UAS) devices it utilizes Asynchronous Notification mechanism to receive events from hardware. Active polling with TEST UNIT READY commands with 3 seconds period is used for incapable hardware. After that both CD and DA drivers work the same way, detecting two conditions: "NOT READY: Medium not present" after medium was detected previously, and "UNIT ATTENTION: Not ready to ready change, medium may have changed". First one reported to disk(9) as media removal, second as media insert/change. To reliably receive second event new AC_UNIT_ATTENTION async added to make UAs broadcasted to all periphs by generic error handling code in cam_periph_error(). 2) Modifications to GEOM core to handle media remove and change events. Media removal handled by spoiling all consumers attached to the provider. Media change event also schedules provider retaste after spoiling to probe new media. New flag G_CF_ORPHAN was added to consumers to reflect that consumer is in process of destruction. It allows retaste to create new geom instance of the same class, while previous one is still dying. 3) Modifications to some GEOM classes: DEV -- to report media change events to devd; VFS -- to handle spoiling same as orphan to prevent accessing replaced media. PART class already handles spoiling alike to orphan. Reviewed by: silence on geom@ and scsi@ Tested by: avg Sponsored by: iXsystems, Inc. / PC-BSD MFC after: 2 months
2012-07-29 11:51:48 +00:00
if (cam_periph_acquire(periph) == CAM_REQ_CMP) {
softc->tur = 1;
daschedule(periph);
}
}
/* Queue us up again */
if (da_poll_period != 0)
callout_schedule(&softc->mediapoll_c, da_poll_period * hz);
}
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
static void
daprevent(struct cam_periph *periph, int action)
{
struct da_softc *softc;
union ccb *ccb;
int error;
softc = (struct da_softc *)periph->softc;
if (((action == PR_ALLOW)
&& (softc->flags & DA_FLAG_PACK_LOCKED) == 0)
|| ((action == PR_PREVENT)
&& (softc->flags & DA_FLAG_PACK_LOCKED) != 0)) {
return;
}
Replace most of priority numbers with defines. No logical changes.
2009-10-23 08:27:55 +00:00
ccb = cam_periph_getccb(periph, CAM_PRIORITY_NORMAL);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
scsi_prevent(&ccb->csio,
/*retries*/1,
/*cbcfp*/dadone,
MSG_SIMPLE_Q_TAG,
action,
SSD_FULL_SIZE,
5000);
Do not reinvent a wheel and let default error handler do its job.
2012-06-05 10:08:22 +00:00
error = cam_periph_runccb(ccb, daerror, CAM_RETRY_SELTO,
Suppress error printing for "PREVENT ALLOW MEDIUM REMOVAL" on da open. Change at r250208 exposed more errors here, hidden before. The same flag is used in cd driver.
2013-05-12 09:33:33 +00:00
SF_RETRY_UA | SF_NO_PRINT, softc->disk->d_devstat);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
if (error == 0) {
if (action == PR_ALLOW)
softc->flags &= ~DA_FLAG_PACK_LOCKED;
else
softc->flags |= DA_FLAG_PACK_LOCKED;
}
xpt_release_ccb(ccb);
}
static void
Merge to da driver quirks hinting 4K physical sector sizes for SATA disks connected via SAS or USB. Unluckily I've found that SAS (mps) and USB-SATA I have translate models in different ways, requiring twice more quirks. Unluckily for Hitachi, their model names are trimmed on SAS, making impossible to identify 4K sector drives that way.
2011-12-23 00:10:17 +00:00
dasetgeom(struct cam_periph *periph, uint32_t block_len, uint64_t maxsector,
Add CAM infrastructure to allow reporting when a drive's long read capacity data changes. cam_ccb.h: Add a new advanced information type, CDAI_TYPE_RCAPLONG, for long read capacity data. cam_xpt_internal.h: Add a read capacity data pointer and length to struct cam_ed. cam_xpt.c: Free the read capacity buffer when a device goes away. While we're here, make sure we don't leak memory for other malloced fields in struct cam_ed. scsi_all.c: Update the scsi_read_capacity_16() to take a uint8_t * and a length instead of just a pointer to the parameter data structure. This will hopefully make this function somewhat immune to future changes in the parameter data. scsi_all.h: Add some extra bit definitions to struct scsi_read_capacity_data_long, and bump up the structure size to the full size specified by SBC-3. Change the prototype for scsi_read_capacity_16(). scsi_da.c: Register changes in read capacity data with the transport layer. This allows the transport layer to send out an async notification to interested parties. Update the dasetgeom() API. Use scsi_extract_sense_len() instead of scsi_extract_sense(). scsi_xpt.c: Add support for the new CDAI_TYPE_RCAPLONG advanced information type. Make sure we set the physpath pointer to NULL after freeing it. This allows blindly freeing it in the struct cam_ed destructor. sys/param.h: Bump __FreeBSD_version from 1000005 to 1000006 to make it easier for third party drivers to determine that the read capacity data async notification is available. camcontrol.c, mptutil/mpt_cam.c: Update these for the new scsi_read_capacity_16() argument structure. Sponsored by: Spectra Logic
2012-01-26 18:09:28 +00:00
struct scsi_read_capacity_data_long *rcaplong, size_t rcap_len)
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
{
struct ccb_calc_geometry ccg;
struct da_softc *softc;
struct disk_params *dp;
Add CAM infrastructure to allow reporting when a drive's long read capacity data changes. cam_ccb.h: Add a new advanced information type, CDAI_TYPE_RCAPLONG, for long read capacity data. cam_xpt_internal.h: Add a read capacity data pointer and length to struct cam_ed. cam_xpt.c: Free the read capacity buffer when a device goes away. While we're here, make sure we don't leak memory for other malloced fields in struct cam_ed. scsi_all.c: Update the scsi_read_capacity_16() to take a uint8_t * and a length instead of just a pointer to the parameter data structure. This will hopefully make this function somewhat immune to future changes in the parameter data. scsi_all.h: Add some extra bit definitions to struct scsi_read_capacity_data_long, and bump up the structure size to the full size specified by SBC-3. Change the prototype for scsi_read_capacity_16(). scsi_da.c: Register changes in read capacity data with the transport layer. This allows the transport layer to send out an async notification to interested parties. Update the dasetgeom() API. Use scsi_extract_sense_len() instead of scsi_extract_sense(). scsi_xpt.c: Add support for the new CDAI_TYPE_RCAPLONG advanced information type. Make sure we set the physpath pointer to NULL after freeing it. This allows blindly freeing it in the struct cam_ed destructor. sys/param.h: Bump __FreeBSD_version from 1000005 to 1000006 to make it easier for third party drivers to determine that the read capacity data async notification is available. camcontrol.c, mptutil/mpt_cam.c: Update these for the new scsi_read_capacity_16() argument structure. Sponsored by: Spectra Logic
2012-01-26 18:09:28 +00:00
u_int lbppbe, lalba;
Fix locking problem in disk_resize(); previously it would run without topology lock, resulting in assertion when running with DIAGNOSTIC. Reviewed by: mav (earlier version)
2012-10-29 17:52:43 +00:00
int error;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
softc = (struct da_softc *)periph->softc;
dp = &softc->params;
Add support to CAM for devices with more than 2^32 blocks. (2TB if you're using 512 byte blocks). cam_ccb.h: Bump up volume_size and cylinders in ccb_calc_geometry to 64 bits and 32 bits respectively, so we can hold larger device sizes. cylinders would overflow at about 500GB. Bump CAM_VERSION for this change. Note that this will require a recompile of all applications that talk to the pass(4) driver. scsi_all.c: Add descriptions for READ/WRITE(16), update READ/WRITE(12) descriptions, add descriptions for SERVICE ACTION IN/OUT. Add a new function, scsi_read_capacity_16(), that issues the read capacity service action. (Necessary for arrays larger than 2^32 sectors.) Update scsi_read_write() to use a 64 bit LBA and issue READ(16) or WRITE(16) if necessary. NOTE the API change. This should be largely transparnet to most userland applications at compile time, but will break binary compatibility. The CAM_VERSION bump, above, also serves the purpose of forcing a recompile for any applications that talk to CAM. scsi_all.h: Add 16 byte READ/WRITE structures, structures for 16 byte READ CAPACITY/SERVICE ACTION IN. Add scsi_u64to8b() and scsi_8btou64. scsi_da.c: The da(4) driver probe now has two stages for devices larger than 2TB. If a standard READ CAPACITY(10) returns 0xffffffff, we issue the 16 byte version of read capacity to determine the true array capacity. We also do the same thing in daopen() -- use the 16 byte read capacity if the device is large enough. The sysctl/loader code has also been updated to accept 16 bytes as a minimum command size.
2003-04-30 00:35:22 +00:00
dp->secsize = block_len;
dp->sectors = maxsector + 1;
Add CAM infrastructure to allow reporting when a drive's long read capacity data changes. cam_ccb.h: Add a new advanced information type, CDAI_TYPE_RCAPLONG, for long read capacity data. cam_xpt_internal.h: Add a read capacity data pointer and length to struct cam_ed. cam_xpt.c: Free the read capacity buffer when a device goes away. While we're here, make sure we don't leak memory for other malloced fields in struct cam_ed. scsi_all.c: Update the scsi_read_capacity_16() to take a uint8_t * and a length instead of just a pointer to the parameter data structure. This will hopefully make this function somewhat immune to future changes in the parameter data. scsi_all.h: Add some extra bit definitions to struct scsi_read_capacity_data_long, and bump up the structure size to the full size specified by SBC-3. Change the prototype for scsi_read_capacity_16(). scsi_da.c: Register changes in read capacity data with the transport layer. This allows the transport layer to send out an async notification to interested parties. Update the dasetgeom() API. Use scsi_extract_sense_len() instead of scsi_extract_sense(). scsi_xpt.c: Add support for the new CDAI_TYPE_RCAPLONG advanced information type. Make sure we set the physpath pointer to NULL after freeing it. This allows blindly freeing it in the struct cam_ed destructor. sys/param.h: Bump __FreeBSD_version from 1000005 to 1000006 to make it easier for third party drivers to determine that the read capacity data async notification is available. camcontrol.c, mptutil/mpt_cam.c: Update these for the new scsi_read_capacity_16() argument structure. Sponsored by: Spectra Logic
2012-01-26 18:09:28 +00:00
if (rcaplong != NULL) {
lbppbe = rcaplong->prot_lbppbe & SRC16_LBPPBE;
lalba = scsi_2btoul(rcaplong->lalba_lbp);
lalba &= SRC16_LALBA_A;
} else {
lbppbe = 0;
lalba = 0;
}
Merge to da driver quirks hinting 4K physical sector sizes for SATA disks connected via SAS or USB. Unluckily I've found that SAS (mps) and USB-SATA I have translate models in different ways, requiring twice more quirks. Unluckily for Hitachi, their model names are trimmed on SAS, making impossible to identify 4K sector drives that way.
2011-12-23 00:10:17 +00:00
if (lbppbe > 0) {
dp->stripesize = block_len << lbppbe;
Report stripeoffset as zero not stripesize if physical block is zero aligned, same as it is done for ATA.
2011-12-23 20:59:13 +00:00
dp->stripeoffset = (dp->stripesize - block_len * lalba) %
dp->stripesize;
Merge to da driver quirks hinting 4K physical sector sizes for SATA disks connected via SAS or USB. Unluckily I've found that SAS (mps) and USB-SATA I have translate models in different ways, requiring twice more quirks. Unluckily for Hitachi, their model names are trimmed on SAS, making impossible to identify 4K sector drives that way.
2011-12-23 00:10:17 +00:00
} else if (softc->quirks & DA_Q_4K) {
dp->stripesize = 4096;
dp->stripeoffset = 0;
} else {
dp->stripesize = 0;
dp->stripeoffset = 0;
}
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
/*
* Have the controller provide us with a geometry
* for this disk. The only time the geometry
* matters is when we boot and the controller
* is the only one knowledgeable enough to come
* up with something that will make this a bootable
* device.
*/
Replace most of priority numbers with defines. No logical changes.
2009-10-23 08:27:55 +00:00
xpt_setup_ccb(&ccg.ccb_h, periph->path, CAM_PRIORITY_NORMAL);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
ccg.ccb_h.func_code = XPT_CALC_GEOMETRY;
ccg.block_size = dp->secsize;
ccg.volume_size = dp->sectors;
ccg.heads = 0;
ccg.secs_per_track = 0;
ccg.cylinders = 0;
xpt_action((union ccb*)&ccg);
Actually check to see if XPT_CALC_GEOMETRY had an error and pick a bogus geometry if it had so we don't later get a divide by zero trap.
2006-04-18 22:01:59 +00:00
if ((ccg.ccb_h.status & CAM_STATUS_MASK) != CAM_REQ_CMP) {
/*
* We don't know what went wrong here- but just pick
* a geometry so we don't have nasty things like divide
* by zero.
*/
dp->heads = 255;
dp->secs_per_track = 255;
dp->cylinders = dp->sectors / (255 * 255);
if (dp->cylinders == 0) {
dp->cylinders = 1;
}
} else {
dp->heads = ccg.heads;
dp->secs_per_track = ccg.secs_per_track;
dp->cylinders = ccg.cylinders;
}
Add CAM infrastructure to allow reporting when a drive's long read capacity data changes. cam_ccb.h: Add a new advanced information type, CDAI_TYPE_RCAPLONG, for long read capacity data. cam_xpt_internal.h: Add a read capacity data pointer and length to struct cam_ed. cam_xpt.c: Free the read capacity buffer when a device goes away. While we're here, make sure we don't leak memory for other malloced fields in struct cam_ed. scsi_all.c: Update the scsi_read_capacity_16() to take a uint8_t * and a length instead of just a pointer to the parameter data structure. This will hopefully make this function somewhat immune to future changes in the parameter data. scsi_all.h: Add some extra bit definitions to struct scsi_read_capacity_data_long, and bump up the structure size to the full size specified by SBC-3. Change the prototype for scsi_read_capacity_16(). scsi_da.c: Register changes in read capacity data with the transport layer. This allows the transport layer to send out an async notification to interested parties. Update the dasetgeom() API. Use scsi_extract_sense_len() instead of scsi_extract_sense(). scsi_xpt.c: Add support for the new CDAI_TYPE_RCAPLONG advanced information type. Make sure we set the physpath pointer to NULL after freeing it. This allows blindly freeing it in the struct cam_ed destructor. sys/param.h: Bump __FreeBSD_version from 1000005 to 1000006 to make it easier for third party drivers to determine that the read capacity data async notification is available. camcontrol.c, mptutil/mpt_cam.c: Update these for the new scsi_read_capacity_16() argument structure. Sponsored by: Spectra Logic
2012-01-26 18:09:28 +00:00
/*
* If the user supplied a read capacity buffer, and if it is
* different than the previous buffer, update the data in the EDT.
* If it's the same, we don't bother. This avoids sending an
* update every time someone opens this device.
*/
if ((rcaplong != NULL)
&& (bcmp(rcaplong, &softc->rcaplong,
min(sizeof(softc->rcaplong), rcap_len)) != 0)) {
struct ccb_dev_advinfo cdai;
xpt_setup_ccb(&cdai.ccb_h, periph->path, CAM_PRIORITY_NORMAL);
cdai.ccb_h.func_code = XPT_DEV_ADVINFO;
cdai.buftype = CDAI_TYPE_RCAPLONG;
Make sure that the flags for the XPT_DEV_ADVINFO CCB are initialized properly. If there is garbage in the flags field, it can sometimes include a set CDAI_FLAG_STORE flag, which may cause either an error or perhaps result in overwriting the field that was intended to be read. sys/cam/cam_ccb.h: Add a new flag to the XPT_DEV_ADVINFO CCB, CDAI_FLAG_NONE, that callers can use to set the flags field when no store is desired. sys/cam/scsi/scsi_enc_ses.c: In ses_setphyspath_callback(), explicitly set the XPT_DEV_ADVINFO flags to CDAI_FLAG_NONE when fetching the physical path information. Instead of ORing in the CDAI_FLAG_STORE flag when storing the physical path, set the flags field to CDAI_FLAG_STORE. sys/cam/scsi/scsi_sa.c: Set the XPT_DEV_ADVINFO flags field to CDAI_FLAG_NONE when fetching extended inquiry information. sys/cam/scsi/scsi_da.c: When storing extended READ CAPACITY information, set the XPT_DEV_ADVINFO flags field to CDAI_FLAG_STORE instead of ORing it into a field that isn't initialized. sys/dev/mpr/mpr_sas.c, sys/dev/mps/mps_sas.c: When fetching extended READ CAPACITY information, set the XPT_DEV_ADVINFO flags field to CDAI_FLAG_NONE instead of setting it to 0. sbin/camcontrol/camcontrol.c: When fetching a device ID, set the XPT_DEV_ADVINFO flags field to CDAI_FLAG_NONE instead of 0. sys/sys/param.h: Bump __FreeBSD_version to 1100061 for the new XPT_DEV_ADVINFO CCB flag, CDAI_FLAG_NONE. Sponsored by: Spectra Logic MFC after: 1 week
2015-02-18 18:30:19 +00:00
cdai.flags = CDAI_FLAG_STORE;
Add CAM infrastructure to allow reporting when a drive's long read capacity data changes. cam_ccb.h: Add a new advanced information type, CDAI_TYPE_RCAPLONG, for long read capacity data. cam_xpt_internal.h: Add a read capacity data pointer and length to struct cam_ed. cam_xpt.c: Free the read capacity buffer when a device goes away. While we're here, make sure we don't leak memory for other malloced fields in struct cam_ed. scsi_all.c: Update the scsi_read_capacity_16() to take a uint8_t * and a length instead of just a pointer to the parameter data structure. This will hopefully make this function somewhat immune to future changes in the parameter data. scsi_all.h: Add some extra bit definitions to struct scsi_read_capacity_data_long, and bump up the structure size to the full size specified by SBC-3. Change the prototype for scsi_read_capacity_16(). scsi_da.c: Register changes in read capacity data with the transport layer. This allows the transport layer to send out an async notification to interested parties. Update the dasetgeom() API. Use scsi_extract_sense_len() instead of scsi_extract_sense(). scsi_xpt.c: Add support for the new CDAI_TYPE_RCAPLONG advanced information type. Make sure we set the physpath pointer to NULL after freeing it. This allows blindly freeing it in the struct cam_ed destructor. sys/param.h: Bump __FreeBSD_version from 1000005 to 1000006 to make it easier for third party drivers to determine that the read capacity data async notification is available. camcontrol.c, mptutil/mpt_cam.c: Update these for the new scsi_read_capacity_16() argument structure. Sponsored by: Spectra Logic
2012-01-26 18:09:28 +00:00
cdai.bufsiz = rcap_len;
cdai.buf = (uint8_t *)rcaplong;
xpt_action((union ccb *)&cdai);
if ((cdai.ccb_h.status & CAM_DEV_QFRZN) != 0)
cam_release_devq(cdai.ccb_h.path, 0, 0, 0, FALSE);
if (cdai.ccb_h.status != CAM_REQ_CMP) {
xpt_print(periph->path, "%s: failed to set read "
"capacity advinfo\n", __func__);
/* Use cam_error_print() to decode the status */
cam_error_print((union ccb *)&cdai, CAM_ESF_CAM_STATUS,
CAM_EPF_ALL);
} else {
bcopy(rcaplong, &softc->rcaplong,
min(sizeof(softc->rcaplong), rcap_len));
}
}
Refactor da(4) to remove one of two code paths used to query capacity data. Reviewed by: ken, mav (earlier version) Sponsored by: The FreeBSD Foundation
2012-04-12 12:58:14 +00:00
softc->disk->d_sectorsize = softc->params.secsize;
softc->disk->d_mediasize = softc->params.secsize * (off_t)softc->params.sectors;
softc->disk->d_stripesize = softc->params.stripesize;
softc->disk->d_stripeoffset = softc->params.stripeoffset;
/* XXX: these are not actually "firmware" values, so they may be wrong */
softc->disk->d_fwsectors = softc->params.secs_per_track;
softc->disk->d_fwheads = softc->params.heads;
softc->disk->d_devstat->block_size = softc->params.secsize;
softc->disk->d_devstat->flags &= ~DEVSTAT_BS_UNAVAILABLE;
Make the da(4) driver notify GEOM about LUN size change. Reviewed by: mav Sponsored by: FreeBSD Foundation
2012-07-07 22:19:51 +00:00
Fix locking problem in disk_resize(); previously it would run without topology lock, resulting in assertion when running with DIAGNOSTIC. Reviewed by: mav (earlier version)
2012-10-29 17:52:43 +00:00
error = disk_resize(softc->disk, M_NOWAIT);
if (error != 0)
xpt_print(periph->path, "disk_resize(9) failed, error = %d\n", error);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
static void
dasendorderedtag(void *arg)
{
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
struct da_softc *softc = arg;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
if (da_send_ordered) {
Some microoptimizations for da and ada drivers: - Replace ordered_tag_count counter with single flag; - From da remove outstanding_cmds counter, duplicating pending_ccbs list; - From da_softc remove unused links field.
2013-10-24 14:05:44 +00:00
if (!LIST_EMPTY(&softc->pending_ccbs)) {
if ((softc->flags & DA_FLAG_WAS_OTAG) == 0)
softc->flags |= DA_FLAG_NEED_OTAG;
softc->flags &= ~DA_FLAG_WAS_OTAG;
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
}
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
/* Queue us up again */
callout_reset(&softc->sendordered_c,
Insert ordered command every 1/4 of the current command timeout, not 1/4 of the default one. Without this change setting kern.cam.ada.default_timeout to 1 instead of 30 allowed me to trigger several false positive command timeouts under heavy ZFS load on a SiI3132 siis(4) controller with 5 HDDs on a port multiplier. MFC after: 1 week
2012-02-02 19:02:15 +00:00
(da_default_timeout * hz) / DA_ORDEREDTAG_INTERVAL,
Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will use to synchornize and protect all data objects that are used for that SIM. Drivers that are not yet MPSAFE register Giant and operate as usual. RIght now, no drivers are MPSAFE, though a few will be changed in the coming week as this work settles down. The driver API has changed, so all CAM drivers will need to be recompiled. The userland API has not changed, so tools like camcontrol do not need to be recompiled.
2007-04-15 08:49:19 +00:00
dasendorderedtag, softc);
SCSI Peripheral drivers for CAM: da - Direct Access Devices (disks, optical devices, SS disks) cd - CDROM (or devices that can act like them, WORM, CD-RW, etc) ch - Medium Changer devices. sa - Sequential Access Devices (tape drives) pass - Application pass-thru driver targ - Target Mode "Processor Target" Emulator pt - Processor Target Devices (scanners, cpus, etc.) Submitted by: The CAM Team
1998-09-15 06:36:34 +00:00
}
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
/*
* Step through all DA peripheral drivers, and if the device is still open,
* sync the disk cache to physical media.
*/
static void
Implement a new generic mechanism for attaching handler functions to events, in order to pave the way for removing a number of the ad-hoc implementations currently in use. Retire the at_shutdown family of functions and replace them with new event handler lists. Rework kern_shutdown.c to take greater advantage of the use of event handlers. Reviewed by: green
1999-08-21 06:24:40 +00:00
dashutdown(void * arg, int howto)
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
{
struct cam_periph *periph;
struct da_softc *softc;
Make pre-shutdown flush and spindown routines to not use xpt_polled_action(), but execute the commands in regular way. There is no any reason to cook CPU while the system is still fully operational. After this change polling in CAM is used only for kernel dumping.
2013-03-29 08:33:18 +00:00
union ccb *ccb;
One more major cam_periph_error() rewrite to improve error handling and reporting. It includes: - removing of error messages controlled by bootverbose, replacing them with more universal and informative debugging on CAM_DEBUG_INFO level, that is now built into the kernel by default; - more close following to the arguments submitted by caller, such as SF_PRINT_ALWAYS, SF_QUIET_IR and SF_NO_PRINT; consumer knows better which errors are usual/expected at this point and which are really informative; - adding two new flags SF_NO_RECOVERY and SF_NO_RETRY to allow caller specify how much assistance it needs at this point; previously consumers controlled that by not calling cam_periph_error() at all, but that made behavior inconsistent and debugging complicated; - tuning debug messages and taken actions order to make debugging output more readable and cause-effect relationships visible; - making camperiphdone() (common device recovery completion handler) to also use cam_periph_error() in most cases, instead of own dumb code; - removing manual sense fetching code from cam_periph_error(); I was told by number of people that it is SIM obligation to fetch sense data, so this code is useless and only significantly complicates recovery logic; - making ada, da and pass driver to use cam_periph_error() with new limited recovery options to handle error recovery and debugging in common way; as one of results, CAM_REQUEUE_REQ and other retrying statuses are now working fine with pass driver, that caused many problems before. - reverting r186891 by raj@ to avoid burning few seconds in tight DELAY() loops on device probe, while device simply loads media; I think that problem may already be fixed in other way, and even if it is not, solution must be different. Sponsored by: iXsystems, Inc. MFC after: 2 weeks
2012-06-09 13:07:44 +00:00
int error;
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
Implement CAM_PERIPH_FOREACH() macro, safely iterating over the list of driver's periphs, acquiring and releaseing periph references while doing it. Use it to iterate over the lists of ada and da periphs when flushing caches and putting devices to sleep on shutdown and suspend. Previous code could panic in theory if some device disappear in the middle of the process.
2013-03-29 07:50:47 +00:00
CAM_PERIPH_FOREACH(periph, &dadriver) {
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
softc = (struct da_softc *)periph->softc;
Restore use of polling mode for disk cache flush in case of kernel panic. While I am not sure that any extra hardware access is a good idea after panic, that is an existing behaviour that should better work correctly.
2013-06-15 12:46:38 +00:00
if (SCHEDULER_STOPPED()) {
/* If we paniced with the lock held, do not recurse. */
if (!cam_periph_owned(periph) &&
(softc->flags & DA_FLAG_OPEN)) {
dadump(softc->disk, NULL, 0, 0, 0);
}
continue;
}
cam_periph_lock(periph);
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
/*
* We only sync the cache if the drive is still open, and
* if the drive is capable of it..
*/
if (((softc->flags & DA_FLAG_OPEN) == 0)
Fix a leaked lock in dashutdown.
2007-04-19 22:18:15 +00:00
|| (softc->quirks & DA_Q_NO_SYNC_CACHE)) {
cam_periph_unlock(periph);
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
continue;
Fix a leaked lock in dashutdown.
2007-04-19 22:18:15 +00:00
}
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
Make pre-shutdown flush and spindown routines to not use xpt_polled_action(), but execute the commands in regular way. There is no any reason to cook CPU while the system is still fully operational. After this change polling in CAM is used only for kernel dumping.
2013-03-29 08:33:18 +00:00
ccb = cam_periph_getccb(periph, CAM_PRIORITY_NORMAL);
scsi_synchronize_cache(&ccb->csio,
One more major cam_periph_error() rewrite to improve error handling and reporting. It includes: - removing of error messages controlled by bootverbose, replacing them with more universal and informative debugging on CAM_DEBUG_INFO level, that is now built into the kernel by default; - more close following to the arguments submitted by caller, such as SF_PRINT_ALWAYS, SF_QUIET_IR and SF_NO_PRINT; consumer knows better which errors are usual/expected at this point and which are really informative; - adding two new flags SF_NO_RECOVERY and SF_NO_RETRY to allow caller specify how much assistance it needs at this point; previously consumers controlled that by not calling cam_periph_error() at all, but that made behavior inconsistent and debugging complicated; - tuning debug messages and taken actions order to make debugging output more readable and cause-effect relationships visible; - making camperiphdone() (common device recovery completion handler) to also use cam_periph_error() in most cases, instead of own dumb code; - removing manual sense fetching code from cam_periph_error(); I was told by number of people that it is SIM obligation to fetch sense data, so this code is useless and only significantly complicates recovery logic; - making ada, da and pass driver to use cam_periph_error() with new limited recovery options to handle error recovery and debugging in common way; as one of results, CAM_REQUEUE_REQ and other retrying statuses are now working fine with pass driver, that caused many problems before. - reverting r186891 by raj@ to avoid burning few seconds in tight DELAY() loops on device probe, while device simply loads media; I think that problem may already be fixed in other way, and even if it is not, solution must be different. Sponsored by: iXsystems, Inc. MFC after: 2 weeks
2012-06-09 13:07:44 +00:00
/*retries*/0,
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
/*cbfcnp*/dadone,
MSG_SIMPLE_Q_TAG,
/*begin_lba*/0, /* whole disk */
/*lb_count*/0,
SSD_FULL_SIZE,
Be very generous with timeouts for synchronize cache. We may wait a very long time in the cases where it really sends the drive out to lunch, but it also allows us to catch very wierd edge cases of strange drives that might take a very long time (emulated disk drives over a network, e.g.).
2001-11-17 18:26:00 +00:00
60 * 60 * 1000);
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
Make pre-shutdown flush and spindown routines to not use xpt_polled_action(), but execute the commands in regular way. There is no any reason to cook CPU while the system is still fully operational. After this change polling in CAM is used only for kernel dumping.
2013-03-29 08:33:18 +00:00
error = cam_periph_runccb(ccb, daerror, /*cam_flags*/0,
/*sense_flags*/ SF_NO_RECOVERY | SF_NO_RETRY | SF_QUIET_IR,
softc->disk->d_devstat);
One more major cam_periph_error() rewrite to improve error handling and reporting. It includes: - removing of error messages controlled by bootverbose, replacing them with more universal and informative debugging on CAM_DEBUG_INFO level, that is now built into the kernel by default; - more close following to the arguments submitted by caller, such as SF_PRINT_ALWAYS, SF_QUIET_IR and SF_NO_PRINT; consumer knows better which errors are usual/expected at this point and which are really informative; - adding two new flags SF_NO_RECOVERY and SF_NO_RETRY to allow caller specify how much assistance it needs at this point; previously consumers controlled that by not calling cam_periph_error() at all, but that made behavior inconsistent and debugging complicated; - tuning debug messages and taken actions order to make debugging output more readable and cause-effect relationships visible; - making camperiphdone() (common device recovery completion handler) to also use cam_periph_error() in most cases, instead of own dumb code; - removing manual sense fetching code from cam_periph_error(); I was told by number of people that it is SIM obligation to fetch sense data, so this code is useless and only significantly complicates recovery logic; - making ada, da and pass driver to use cam_periph_error() with new limited recovery options to handle error recovery and debugging in common way; as one of results, CAM_REQUEUE_REQ and other retrying statuses are now working fine with pass driver, that caused many problems before. - reverting r186891 by raj@ to avoid burning few seconds in tight DELAY() loops on device probe, while device simply loads media; I think that problem may already be fixed in other way, and even if it is not, solution must be different. Sponsored by: iXsystems, Inc. MFC after: 2 weeks
2012-06-09 13:07:44 +00:00
if (error != 0)
xpt_print(periph->path, "Synchronize cache failed\n");
Add xpt_release_ccb()'s missed at r248872. That made `shutdown -p` stuck on controller with small number of queue slots and several disks connected.
2013-04-03 11:30:18 +00:00
xpt_release_ccb(ccb);
Missed locking the dump and shutdown entry points in the scsi_da driver.
2007-04-18 05:14:16 +00:00
cam_periph_unlock(periph);
Add the quirk entry framework to handle disabling the synchronize cache command on drives that don't like it. Right now, there's just a bogus quirk entry in the table that doesn't do anything, but that should be changed once we get actual inquiry data for drives that don't like the synchronize cache command. Also, add a shutdown hook that runs through all direct access peripherals and runs a synchronize cache on them if they're still open, and if synchronize cache isn't disabled via a quirk entry. Add a synchronize cache call at the end of dadump() (again, conditionalized on the quirk entry), so we can insure that the disk cache contents get flushed to physical media after a dump. Check the new quirk entry in daclose() to decide whether or not to synchronize the cache for a disk at final close. Reviewed by: gibbs
1998-10-08 05:46:38 +00:00
}
}
Implement a new camcontrol function, 'camcontrol format'. libcam/Makefile: Add scsi_da.c to libcam for the new scsi_format_unit() function. camcontrol.8: Update the man page for the new format functionality, and take out the examples section describing how to do it with 'camcontrol cmd'. camcontrol.c: New format functionality. Note that unlike the rest of the camcontrol subcommands, this one is interactive by default. Because of the potential destructiveness of the format command, I thought it necessary to get confirmation from the user before spamming a disk. You can disable the interactive behavior, and the status meter with command line arguments. scsi_da.c: Add the new scsi_format_unit() cdb building function and use #ifdef _KERNEL to make this file compile in both the kernel and userland. The format unit function is currently only defined in the non-kernel case, because nothing in the kernel is using it. If that changes, it should be un-ifdefed and compiled in both cases. scsi_da.h: New function declaration, CDB structure and format data structures. Thanks to Nick Hibma for providing some valuable input on these changes.
2000-05-21 23:57:52 +00:00
#else /* !_KERNEL */
/*
Improve camcontrol(8) handling of drive defect data. This includes a new summary mode (-s) for camcontrol defects that quickly tells the user the most important thing: how many defects are in the requested list. The actual location of the defects is less important. Modern drives frequently have more than the 8191 defects that can be reported by the READ DEFECT DATA (10) command. If they don't have that many grown defects, they certainly have more than 8191 defects in the primary (i.e. factory) defect list. The READ DEFECT DATA (12) command allows for longer parameter lists, as well as indexing into the list of defects, and so allows reporting many more defects. This has been tested with HGST drives and Seagate drives, but does not fully work with Seagate drives. Once I have a Seagate spec I may be able to determine whether it is possible to make it work with Seagate drives. scsi_da.h: Add a definition for the new long block defect format. Add bit and mask definitions for the new extended physical sector and bytes from index defect formats. Add a prototype for the new scsi_read_defects() CDB building function. scsi_da.c: Add a new scsi_read_defects() CDB building function. camcontrol(8) was previously composing CDBs manually. This is long overdue. camcontrol.c: Revamp the camcontrol defects subcommand. We now go through multiple stages in trying to get defect data off the drive while avoiding various drive firmware quirks. We start off by requesting the defect header with the 10 byte command. If we're in summary mode (-s) and the drive reports fewer defects than can be represented in the 10 byte header, we're done. Otherwise, we know that we need to issue the 12 byte command if the drive reports the maximum number of defects. If we're in summary mode, we're done if we get a good response back when asking for the 12 byte header. If the user has asked for the full list, then we use the address descriptor index field in the 12 byte CDB to step through the list in 64K chunks. 64K is small enough to work with most any ancient or modern SCSI controller. Add support for printing the new long block defect format, as well as the extended physical sector and bytes from index formats. I don't have any drives that support the new formats. Add a hexadecimal output format that can be turned on with -X. Add a quiet mode (-q) that can be turned on with the summary mode (-s) to just print out a number. Revamp the error detection and recovery code for the defects command to work with HGST drives. Call the new scsi_read_defects() CDB building function instead of rolling the CDB ourselves. Pay attention to the residual from the defect list request when printing it out, so we don't run off the end of the list. Use the new scsi_nv library routines to convert from strings to numbers and back. camcontrol.8: Document the new defect formats (longblock, extbfi, extphys) and command line options (-q, -s, -S and -X) for the defects subcommand. Explain a little more about what drives generally do and don't support. Sponsored by: Spectra Logic MFC after: 1 week
2015-01-08 16:58:40 +00:00
* XXX These are only left out of the kernel build to silence warnings. If,
* for some reason these functions are used in the kernel, the ifdefs should
* be moved so they are included both in the kernel and userland.
Implement a new camcontrol function, 'camcontrol format'. libcam/Makefile: Add scsi_da.c to libcam for the new scsi_format_unit() function. camcontrol.8: Update the man page for the new format functionality, and take out the examples section describing how to do it with 'camcontrol cmd'. camcontrol.c: New format functionality. Note that unlike the rest of the camcontrol subcommands, this one is interactive by default. Because of the potential destructiveness of the format command, I thought it necessary to get confirmation from the user before spamming a disk. You can disable the interactive behavior, and the status meter with command line arguments. scsi_da.c: Add the new scsi_format_unit() cdb building function and use #ifdef _KERNEL to make this file compile in both the kernel and userland. The format unit function is currently only defined in the non-kernel case, because nothing in the kernel is using it. If that changes, it should be un-ifdefed and compiled in both cases. scsi_da.h: New function declaration, CDB structure and format data structures. Thanks to Nick Hibma for providing some valuable input on these changes.
2000-05-21 23:57:52 +00:00
*/
void
scsi_format_unit(struct ccb_scsiio *csio, u_int32_t retries,
void (*cbfcnp)(struct cam_periph *, union ccb *),
u_int8_t tag_action, u_int8_t byte2, u_int16_t ileave,
u_int8_t *data_ptr, u_int32_t dxfer_len, u_int8_t sense_len,
u_int32_t timeout)
{
struct scsi_format_unit *scsi_cmd;
scsi_cmd = (struct scsi_format_unit *)&csio->cdb_io.cdb_bytes;
scsi_cmd->opcode = FORMAT_UNIT;
scsi_cmd->byte2 = byte2;
scsi_ulto2b(ileave, scsi_cmd->interleave);
cam_fill_csio(csio,
retries,
cbfcnp,
/*flags*/ (dxfer_len > 0) ? CAM_DIR_OUT : CAM_DIR_NONE,
tag_action,
data_ptr,
dxfer_len,
sense_len,
sizeof(*scsi_cmd),
timeout);
}
Improve camcontrol(8) handling of drive defect data. This includes a new summary mode (-s) for camcontrol defects that quickly tells the user the most important thing: how many defects are in the requested list. The actual location of the defects is less important. Modern drives frequently have more than the 8191 defects that can be reported by the READ DEFECT DATA (10) command. If they don't have that many grown defects, they certainly have more than 8191 defects in the primary (i.e. factory) defect list. The READ DEFECT DATA (12) command allows for longer parameter lists, as well as indexing into the list of defects, and so allows reporting many more defects. This has been tested with HGST drives and Seagate drives, but does not fully work with Seagate drives. Once I have a Seagate spec I may be able to determine whether it is possible to make it work with Seagate drives. scsi_da.h: Add a definition for the new long block defect format. Add bit and mask definitions for the new extended physical sector and bytes from index defect formats. Add a prototype for the new scsi_read_defects() CDB building function. scsi_da.c: Add a new scsi_read_defects() CDB building function. camcontrol(8) was previously composing CDBs manually. This is long overdue. camcontrol.c: Revamp the camcontrol defects subcommand. We now go through multiple stages in trying to get defect data off the drive while avoiding various drive firmware quirks. We start off by requesting the defect header with the 10 byte command. If we're in summary mode (-s) and the drive reports fewer defects than can be represented in the 10 byte header, we're done. Otherwise, we know that we need to issue the 12 byte command if the drive reports the maximum number of defects. If we're in summary mode, we're done if we get a good response back when asking for the 12 byte header. If the user has asked for the full list, then we use the address descriptor index field in the 12 byte CDB to step through the list in 64K chunks. 64K is small enough to work with most any ancient or modern SCSI controller. Add support for printing the new long block defect format, as well as the extended physical sector and bytes from index formats. I don't have any drives that support the new formats. Add a hexadecimal output format that can be turned on with -X. Add a quiet mode (-q) that can be turned on with the summary mode (-s) to just print out a number. Revamp the error detection and recovery code for the defects command to work with HGST drives. Call the new scsi_read_defects() CDB building function instead of rolling the CDB ourselves. Pay attention to the residual from the defect list request when printing it out, so we don't run off the end of the list. Use the new scsi_nv library routines to convert from strings to numbers and back. camcontrol.8: Document the new defect formats (longblock, extbfi, extphys) and command line options (-q, -s, -S and -X) for the defects subcommand. Explain a little more about what drives generally do and don't support. Sponsored by: Spectra Logic MFC after: 1 week
2015-01-08 16:58:40 +00:00
void
scsi_read_defects(struct ccb_scsiio *csio, uint32_t retries,
void (*cbfcnp)(struct cam_periph *, union ccb *),
uint8_t tag_action, uint8_t list_format,
uint32_t addr_desc_index, uint8_t *data_ptr,
uint32_t dxfer_len, int minimum_cmd_size,
uint8_t sense_len, uint32_t timeout)
{
uint8_t cdb_len;
/*
* These conditions allow using the 10 byte command. Otherwise we
* need to use the 12 byte command.
*/
if ((minimum_cmd_size <= 10)
&& (addr_desc_index == 0)
&& (dxfer_len <= SRDD10_MAX_LENGTH)) {
struct scsi_read_defect_data_10 *cdb10;
cdb10 = (struct scsi_read_defect_data_10 *)
&csio->cdb_io.cdb_bytes;
cdb_len = sizeof(*cdb10);
bzero(cdb10, cdb_len);
cdb10->opcode = READ_DEFECT_DATA_10;
cdb10->format = list_format;
scsi_ulto2b(dxfer_len, cdb10->alloc_length);
} else {
struct scsi_read_defect_data_12 *cdb12;
cdb12 = (struct scsi_read_defect_data_12 *)
&csio->cdb_io.cdb_bytes;
cdb_len = sizeof(*cdb12);
bzero(cdb12, cdb_len);
cdb12->opcode = READ_DEFECT_DATA_12;
cdb12->format = list_format;
scsi_ulto4b(dxfer_len, cdb12->alloc_length);
scsi_ulto4b(addr_desc_index, cdb12->address_descriptor_index);
}
cam_fill_csio(csio,
retries,
cbfcnp,
/*flags*/ CAM_DIR_IN,
tag_action,
data_ptr,
dxfer_len,
sense_len,
cdb_len,
timeout);
}
Add camcontrol support for the SCSI sanitize command Reviewed by: ken, mjacob (eariler version) Sponsored by: Netapp
2013-09-06 15:19:57 +00:00
void
scsi_sanitize(struct ccb_scsiio *csio, u_int32_t retries,
void (*cbfcnp)(struct cam_periph *, union ccb *),
u_int8_t tag_action, u_int8_t byte2, u_int16_t control,
u_int8_t *data_ptr, u_int32_t dxfer_len, u_int8_t sense_len,
u_int32_t timeout)
{
struct scsi_sanitize *scsi_cmd;
scsi_cmd = (struct scsi_sanitize *)&csio->cdb_io.cdb_bytes;
scsi_cmd->opcode = SANITIZE;
scsi_cmd->byte2 = byte2;
scsi_cmd->control = control;
scsi_ulto2b(dxfer_len, scsi_cmd->length);
cam_fill_csio(csio,
retries,
cbfcnp,
/*flags*/ (dxfer_len > 0) ? CAM_DIR_OUT : CAM_DIR_NONE,
tag_action,
data_ptr,
dxfer_len,
sense_len,
sizeof(*scsi_cmd),
timeout);
}
Implement a new camcontrol function, 'camcontrol format'. libcam/Makefile: Add scsi_da.c to libcam for the new scsi_format_unit() function. camcontrol.8: Update the man page for the new format functionality, and take out the examples section describing how to do it with 'camcontrol cmd'. camcontrol.c: New format functionality. Note that unlike the rest of the camcontrol subcommands, this one is interactive by default. Because of the potential destructiveness of the format command, I thought it necessary to get confirmation from the user before spamming a disk. You can disable the interactive behavior, and the status meter with command line arguments. scsi_da.c: Add the new scsi_format_unit() cdb building function and use #ifdef _KERNEL to make this file compile in both the kernel and userland. The format unit function is currently only defined in the non-kernel case, because nothing in the kernel is using it. If that changes, it should be un-ifdefed and compiled in both cases. scsi_da.h: New function declaration, CDB structure and format data structures. Thanks to Nick Hibma for providing some valuable input on these changes.
2000-05-21 23:57:52 +00:00
#endif /* _KERNEL */
Reference in New Issue Copy Permalink