d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
d9802deb4e
freebsd-skq/sys/dev/mps/mps.c

2683 lines
78 KiB
C
Raw Normal View History

MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
/*-
* Copyright (c) 2009 Yahoo! Inc.
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
* Copyright (c) 2012 LSI Corp.
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE 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.
*
* LSI MPT-Fusion Host Adapter FreeBSD
*
* $FreeBSD$
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
/* Communications core for LSI MPT2 */
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
/* TODO Move headers to mpsvar */
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
#include <sys/types.h>
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/selinfo.h>
#include <sys/lock.h>
#include <sys/mutex.h>
#include <sys/module.h>
#include <sys/bus.h>
#include <sys/conf.h>
#include <sys/bio.h>
#include <sys/malloc.h>
#include <sys/uio.h>
#include <sys/sysctl.h>
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
#include <sys/queue.h>
#include <sys/kthread.h>
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
#include <sys/taskqueue.h>
Fix an event handling bug with the mps(4) driver. This bug manifested itself after repeated device arrivals and departures. The root of the problem was that the last entry in the reply array wasn't initialized/allocated. So every time we got around to that event, we had a bogus address. There were a couple more problems with the code that are also fixed: - The reply mechanism was being treated as sequential (indexed by sc->replycurindex) even though the spec says that the driver should use the ReplyFrameAddress field of the post queue descriptor to figure out where the reply is. There is no guarantee that the reply descriptors will be used in sequential order. - The second word of the reply post queue descriptor wasn't being checked in mps_intr_locked() to make sure that it wasn't 0xffffffff. So the driver could potentially come across a partially DMAed descriptor. - The number of replies allocated was one less than the actual size of the queue. Instead, it was the size of the number of replies that can be used at one time. (Which is one less than the size of the queue.) mps.c: When initializing the entries in the reply free queue, make sure we initialize the full number that we tell the chip we have (sc->fqdepth), not the number that can be used at any one time (sc->num_replies). When allocating replies, make sure we allocate the number of replies that we've told the chip exist, not just the number that can be used simultaneously. Use the ReplyFrameAddress field of the post queue descriptor to figure out which reply is being referenced. This is what the spec says to do, and the spec doesn't guarantee that the replies will be used in order. Put a check in to verify that the reply address passed back from the card is valid. (Panic if it isn't, we'll panic when we try to deference the reply pointer in any case.) In mps_intr_locked(), verify that the second word of the post queue descriptor is not 0xffffffff in addition to verifying that the unused flag is not set, so we can make sure we didn't get a partially DMAed descriptor. Remove references to sc->replycurindex, it isn't needed now. mpsvar.h: Remove replycurindex from the softc, it isn't needed now. Reviewed by: scottl
2010-12-10 21:45:10 +00:00
#include <sys/endian.h>
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
#include <sys/eventhandler.h>
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
#include <machine/bus.h>
#include <machine/resource.h>
#include <sys/rman.h>
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
#include <sys/proc.h>
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
#include <dev/pci/pcivar.h>
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
#include <cam/cam.h>
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
#include <cam/scsi/scsi_all.h>
#include <dev/mps/mpi/mpi2_type.h>
#include <dev/mps/mpi/mpi2.h>
#include <dev/mps/mpi/mpi2_ioc.h>
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
#include <dev/mps/mpi/mpi2_sas.h>
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
#include <dev/mps/mpi/mpi2_cnfg.h>
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
#include <dev/mps/mpi/mpi2_init.h>
#include <dev/mps/mpi/mpi2_tool.h>
#include <dev/mps/mps_ioctl.h>
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
#include <dev/mps/mpsvar.h>
#include <dev/mps/mps_table.h>
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
#include <dev/mps/mps_sas.h>
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
static int mps_diag_reset(struct mps_softc *sc, int sleep_flag);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
static int mps_init_queues(struct mps_softc *sc);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
static int mps_message_unit_reset(struct mps_softc *sc, int sleep_flag);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
static int mps_transition_operational(struct mps_softc *sc);
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
static int mps_iocfacts_allocate(struct mps_softc *sc, uint8_t attaching);
static void mps_iocfacts_free(struct mps_softc *sc);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
static void mps_startup(void *arg);
static int mps_send_iocinit(struct mps_softc *sc);
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
static int mps_alloc_queues(struct mps_softc *sc);
static int mps_alloc_replies(struct mps_softc *sc);
static int mps_alloc_requests(struct mps_softc *sc);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
static int mps_attach_log(struct mps_softc *sc);
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
static __inline void mps_complete_command(struct mps_softc *sc,
struct mps_command *cm);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
static void mps_dispatch_event(struct mps_softc *sc, uintptr_t data,
MPI2_EVENT_NOTIFICATION_REPLY *reply);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
static void mps_config_complete(struct mps_softc *sc, struct mps_command *cm);
static void mps_periodic(void *);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
static int mps_reregister_events(struct mps_softc *sc);
static void mps_enqueue_request(struct mps_softc *sc, struct mps_command *cm);
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
static int mps_get_iocfacts(struct mps_softc *sc, MPI2_IOC_FACTS_REPLY *facts);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
static int mps_wait_db_ack(struct mps_softc *sc, int timeout, int sleep_flag);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
SYSCTL_NODE(_hw, OID_AUTO, mps, CTLFLAG_RD, 0, "MPS Driver Parameters");
MALLOC_DEFINE(M_MPT2, "mps", "mpt2 driver memory");
/*
* Do a "Diagnostic Reset" aka a hard reset. This should get the chip out of
* any state and back to its initialization state machine.
*/
static char mpt2_reset_magic[] = { 0x00, 0x0f, 0x04, 0x0b, 0x02, 0x07, 0x0d };
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
/* Added this union to smoothly convert le64toh cm->cm_desc.Words.
* Compiler only support unint64_t to be passed as argument.
* Otherwise it will through below error
* "aggregate value used where an integer was expected"
*/
typedef union _reply_descriptor {
u64 word;
struct {
u32 low;
u32 high;
} u;
}reply_descriptor,address_descriptor;
/*
* sleep_flag can be either CAN_SLEEP or NO_SLEEP.
* If this function is called from process context, it can sleep
* and there is no harm to sleep, in case if this fuction is called
* from Interrupt handler, we can not sleep and need NO_SLEEP flag set.
* based on sleep flags driver will call either msleep, pause or DELAY.
* msleep and pause are of same variant, but pause is used when mps_mtx
* is not hold by driver.
*
*/
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
static int
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
mps_diag_reset(struct mps_softc *sc,int sleep_flag)
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
{
uint32_t reg;
int i, error, tries = 0;
mps_dprint(sc, MPS_TRACE, "%s\n", __func__);
/* Clear any pending interrupts */
mps_regwrite(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET, 0x0);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
/*Force NO_SLEEP for threads prohibited to sleep
* e.a Thread from interrupt handler are prohibited to sleep.
Replace the TDP_NOSLEEPING flag with a counter so that the THREAD_NO_SLEEPING() and THREAD_SLEEPING_OK() macros can nest. Reviewed by: attilio
2013-03-01 22:03:31 +00:00
*/
if (curthread->td_no_sleeping != 0)
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
sleep_flag = NO_SLEEP;
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
/* Push the magic sequence */
error = ETIMEDOUT;
while (tries++ < 20) {
for (i = 0; i < sizeof(mpt2_reset_magic); i++)
mps_regwrite(sc, MPI2_WRITE_SEQUENCE_OFFSET,
mpt2_reset_magic[i]);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
/* wait 100 msec */
if (mtx_owned(&sc->mps_mtx) && sleep_flag == CAN_SLEEP)
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
msleep(&sc->msleep_fake_chan, &sc->mps_mtx, 0,
"mpsdiag", hz/10);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
else if (sleep_flag == CAN_SLEEP)
pause("mpsdiag", hz/10);
else
DELAY(100 * 1000);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
reg = mps_regread(sc, MPI2_HOST_DIAGNOSTIC_OFFSET);
if (reg & MPI2_DIAG_DIAG_WRITE_ENABLE) {
error = 0;
break;
}
}
if (error)
return (error);
/* Send the actual reset. XXX need to refresh the reg? */
mps_regwrite(sc, MPI2_HOST_DIAGNOSTIC_OFFSET,
reg | MPI2_DIAG_RESET_ADAPTER);
/* Wait up to 300 seconds in 50ms intervals */
error = ETIMEDOUT;
for (i = 0; i < 60000; i++) {
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
/* wait 50 msec */
if (mtx_owned(&sc->mps_mtx) && sleep_flag == CAN_SLEEP)
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
msleep(&sc->msleep_fake_chan, &sc->mps_mtx, 0,
"mpsdiag", hz/20);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
else if (sleep_flag == CAN_SLEEP)
pause("mpsdiag", hz/20);
else
DELAY(50 * 1000);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
reg = mps_regread(sc, MPI2_DOORBELL_OFFSET);
if ((reg & MPI2_IOC_STATE_MASK) != MPI2_IOC_STATE_RESET) {
error = 0;
break;
}
}
if (error)
return (error);
mps_regwrite(sc, MPI2_WRITE_SEQUENCE_OFFSET, 0x0);
return (0);
}
static int
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
mps_message_unit_reset(struct mps_softc *sc, int sleep_flag)
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
{
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
MPS_FUNCTRACE(sc);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
mps_regwrite(sc, MPI2_DOORBELL_OFFSET,
MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET <<
MPI2_DOORBELL_FUNCTION_SHIFT);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
if (mps_wait_db_ack(sc, 5, sleep_flag) != 0) {
mps_dprint(sc, MPS_FAULT, "Doorbell handshake failed : <%s>\n",
__func__);
return (ETIMEDOUT);
}
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
return (0);
}
static int
mps_transition_ready(struct mps_softc *sc)
{
uint32_t reg, state;
int error, tries = 0;
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
int sleep_flags;
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
MPS_FUNCTRACE(sc);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
/* If we are in attach call, do not sleep */
sleep_flags = (sc->mps_flags & MPS_FLAGS_ATTACH_DONE)
? CAN_SLEEP:NO_SLEEP;
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
error = 0;
while (tries++ < 5) {
reg = mps_regread(sc, MPI2_DOORBELL_OFFSET);
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_INIT, "Doorbell= 0x%x\n", reg);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
/*
* Ensure the IOC is ready to talk. If it's not, try
* resetting it.
*/
if (reg & MPI2_DOORBELL_USED) {
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
mps_diag_reset(sc, sleep_flags);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
DELAY(50000);
continue;
}
/* Is the adapter owned by another peer? */
if ((reg & MPI2_DOORBELL_WHO_INIT_MASK) ==
(MPI2_WHOINIT_PCI_PEER << MPI2_DOORBELL_WHO_INIT_SHIFT)) {
device_printf(sc->mps_dev, "IOC is under the control "
"of another peer host, aborting initialization.\n");
return (ENXIO);
}
state = reg & MPI2_IOC_STATE_MASK;
if (state == MPI2_IOC_STATE_READY) {
/* Ready to go! */
error = 0;
break;
} else if (state == MPI2_IOC_STATE_FAULT) {
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_FAULT, "IOC in fault state 0x%x, resetting\n",
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
state & MPI2_DOORBELL_FAULT_CODE_MASK);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
mps_diag_reset(sc, sleep_flags);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
} else if (state == MPI2_IOC_STATE_OPERATIONAL) {
/* Need to take ownership */
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
mps_message_unit_reset(sc, sleep_flags);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
} else if (state == MPI2_IOC_STATE_RESET) {
/* Wait a bit, IOC might be in transition */
mps_dprint(sc, MPS_FAULT,
"IOC in unexpected reset state\n");
} else {
mps_dprint(sc, MPS_FAULT,
"IOC in unknown state 0x%x\n", state);
error = EINVAL;
break;
}
/* Wait 50ms for things to settle down. */
DELAY(50000);
}
if (error)
device_printf(sc->mps_dev, "Cannot transition IOC to ready\n");
return (error);
}
static int
mps_transition_operational(struct mps_softc *sc)
{
uint32_t reg, state;
int error;
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
MPS_FUNCTRACE(sc);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
error = 0;
reg = mps_regread(sc, MPI2_DOORBELL_OFFSET);
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_INIT, "Doorbell= 0x%x\n", reg);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
state = reg & MPI2_IOC_STATE_MASK;
if (state != MPI2_IOC_STATE_READY) {
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
if ((error = mps_transition_ready(sc)) != 0) {
mps_dprint(sc, MPS_FAULT,
"%s failed to transition ready\n", __func__);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
return (error);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
}
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
}
error = mps_send_iocinit(sc);
return (error);
}
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
/*
* This is called during attach and when re-initializing due to a Diag Reset.
* IOC Facts is used to allocate many of the structures needed by the driver.
* If called from attach, de-allocation is not required because the driver has
* not allocated any structures yet, but if called from a Diag Reset, previously
* allocated structures based on IOC Facts will need to be freed and re-
* allocated bases on the latest IOC Facts.
*/
static int
mps_iocfacts_allocate(struct mps_softc *sc, uint8_t attaching)
{
int error, i;
Mpi2IOCFactsReply_t saved_facts;
uint8_t saved_mode, reallocating;
struct mpssas_lun *lun, *lun_tmp;
struct mpssas_target *targ;
mps_dprint(sc, MPS_TRACE, "%s\n", __func__);
/* Save old IOC Facts and then only reallocate if Facts have changed */
if (!attaching) {
bcopy(sc->facts, &saved_facts, sizeof(MPI2_IOC_FACTS_REPLY));
}
/*
* Get IOC Facts. In all cases throughout this function, panic if doing
* a re-initialization and only return the error if attaching so the OS
* can handle it.
*/
if ((error = mps_get_iocfacts(sc, sc->facts)) != 0) {
if (attaching) {
mps_dprint(sc, MPS_FAULT, "%s failed to get IOC Facts "
"with error %d\n", __func__, error);
return (error);
} else {
panic("%s failed to get IOC Facts with error %d\n",
__func__, error);
}
}
mps_print_iocfacts(sc, sc->facts);
snprintf(sc->fw_version, sizeof(sc->fw_version),
"%02d.%02d.%02d.%02d",
sc->facts->FWVersion.Struct.Major,
sc->facts->FWVersion.Struct.Minor,
sc->facts->FWVersion.Struct.Unit,
sc->facts->FWVersion.Struct.Dev);
mps_printf(sc, "Firmware: %s, Driver: %s\n", sc->fw_version,
MPS_DRIVER_VERSION);
mps_printf(sc, "IOCCapabilities: %b\n", sc->facts->IOCCapabilities,
"\20" "\3ScsiTaskFull" "\4DiagTrace" "\5SnapBuf" "\6ExtBuf"
"\7EEDP" "\10BiDirTarg" "\11Multicast" "\14TransRetry" "\15IR"
"\16EventReplay" "\17RaidAccel" "\20MSIXIndex" "\21HostDisc");
/*
* If the chip doesn't support event replay then a hard reset will be
* required to trigger a full discovery. Do the reset here then
* retransition to Ready. A hard reset might have already been done,
* but it doesn't hurt to do it again. Only do this if attaching, not
* for a Diag Reset.
*/
if (attaching) {
if ((sc->facts->IOCCapabilities &
MPI2_IOCFACTS_CAPABILITY_EVENT_REPLAY) == 0) {
mps_diag_reset(sc, NO_SLEEP);
if ((error = mps_transition_ready(sc)) != 0) {
mps_dprint(sc, MPS_FAULT, "%s failed to "
"transition to ready with error %d\n",
__func__, error);
return (error);
}
}
}
/*
* Set flag if IR Firmware is loaded. If the RAID Capability has
* changed from the previous IOC Facts, log a warning, but only if
* checking this after a Diag Reset and not during attach.
*/
saved_mode = sc->ir_firmware;
if (sc->facts->IOCCapabilities &
MPI2_IOCFACTS_CAPABILITY_INTEGRATED_RAID)
sc->ir_firmware = 1;
if (!attaching) {
if (sc->ir_firmware != saved_mode) {
mps_dprint(sc, MPS_FAULT, "%s new IR/IT mode in IOC "
"Facts does not match previous mode\n", __func__);
}
}
/* Only deallocate and reallocate if relevant IOC Facts have changed */
reallocating = FALSE;
if ((!attaching) &&
((saved_facts.MsgVersion != sc->facts->MsgVersion) ||
(saved_facts.HeaderVersion != sc->facts->HeaderVersion) ||
(saved_facts.MaxChainDepth != sc->facts->MaxChainDepth) ||
(saved_facts.RequestCredit != sc->facts->RequestCredit) ||
(saved_facts.ProductID != sc->facts->ProductID) ||
(saved_facts.IOCCapabilities != sc->facts->IOCCapabilities) ||
(saved_facts.IOCRequestFrameSize !=
sc->facts->IOCRequestFrameSize) ||
(saved_facts.MaxTargets != sc->facts->MaxTargets) ||
(saved_facts.MaxSasExpanders != sc->facts->MaxSasExpanders) ||
(saved_facts.MaxEnclosures != sc->facts->MaxEnclosures) ||
(saved_facts.HighPriorityCredit != sc->facts->HighPriorityCredit) ||
(saved_facts.MaxReplyDescriptorPostQueueDepth !=
sc->facts->MaxReplyDescriptorPostQueueDepth) ||
(saved_facts.ReplyFrameSize != sc->facts->ReplyFrameSize) ||
(saved_facts.MaxVolumes != sc->facts->MaxVolumes) ||
(saved_facts.MaxPersistentEntries !=
sc->facts->MaxPersistentEntries))) {
reallocating = TRUE;
}
/*
* Some things should be done if attaching or re-allocating after a Diag
* Reset, but are not needed after a Diag Reset if the FW has not
* changed.
*/
if (attaching || reallocating) {
/*
* Check if controller supports FW diag buffers and set flag to
* enable each type.
*/
if (sc->facts->IOCCapabilities &
MPI2_IOCFACTS_CAPABILITY_DIAG_TRACE_BUFFER)
sc->fw_diag_buffer_list[MPI2_DIAG_BUF_TYPE_TRACE].
enabled = TRUE;
if (sc->facts->IOCCapabilities &
MPI2_IOCFACTS_CAPABILITY_SNAPSHOT_BUFFER)
sc->fw_diag_buffer_list[MPI2_DIAG_BUF_TYPE_SNAPSHOT].
enabled = TRUE;
if (sc->facts->IOCCapabilities &
MPI2_IOCFACTS_CAPABILITY_EXTENDED_BUFFER)
sc->fw_diag_buffer_list[MPI2_DIAG_BUF_TYPE_EXTENDED].
enabled = TRUE;
/*
* Set flag if EEDP is supported and if TLR is supported.
*/
if (sc->facts->IOCCapabilities & MPI2_IOCFACTS_CAPABILITY_EEDP)
sc->eedp_enabled = TRUE;
if (sc->facts->IOCCapabilities & MPI2_IOCFACTS_CAPABILITY_TLR)
sc->control_TLR = TRUE;
/*
* Size the queues. Since the reply queues always need one free
* entry, we'll just deduct one reply message here.
*/
sc->num_reqs = MIN(MPS_REQ_FRAMES, sc->facts->RequestCredit);
sc->num_replies = MIN(MPS_REPLY_FRAMES + MPS_EVT_REPLY_FRAMES,
sc->facts->MaxReplyDescriptorPostQueueDepth) - 1;
/*
* Initialize all Tail Queues
*/
TAILQ_INIT(&sc->req_list);
TAILQ_INIT(&sc->high_priority_req_list);
TAILQ_INIT(&sc->chain_list);
TAILQ_INIT(&sc->tm_list);
}
/*
* If doing a Diag Reset and the FW is significantly different
* (reallocating will be set above in IOC Facts comparison), then all
* buffers based on the IOC Facts will need to be freed before they are
* reallocated.
*/
if (reallocating) {
mps_iocfacts_free(sc);
/*
* The number of targets is based on IOC Facts, so free all of
* the allocated LUNs for each target and then the target buffer
* itself.
*/
for (i=0; i< saved_facts.MaxTargets; i++) {
targ = &sc->sassc->targets[i];
SLIST_FOREACH_SAFE(lun, &targ->luns, lun_link,
lun_tmp) {
free(lun, M_MPT2);
}
}
free(sc->sassc->targets, M_MPT2);
sc->sassc->targets = malloc(sizeof(struct mpssas_target) *
sc->facts->MaxTargets, M_MPT2, M_WAITOK|M_ZERO);
if (!sc->sassc->targets) {
panic("%s failed to alloc targets with error %d\n",
__func__, ENOMEM);
}
}
/*
* Any deallocation has been completed. Now start reallocating
* if needed. Will only need to reallocate if attaching or if the new
* IOC Facts are different from the previous IOC Facts after a Diag
* Reset. Targets have already been allocated above if needed.
*/
if (attaching || reallocating) {
if (((error = mps_alloc_queues(sc)) != 0) ||
((error = mps_alloc_replies(sc)) != 0) ||
((error = mps_alloc_requests(sc)) != 0)) {
if (attaching ) {
mps_dprint(sc, MPS_FAULT, "%s failed to alloc "
"queues with error %d\n", __func__, error);
mps_free(sc);
return (error);
} else {
panic("%s failed to alloc queues with error "
"%d\n", __func__, error);
}
}
}
/* Always initialize the queues */
bzero(sc->free_queue, sc->fqdepth * 4);
mps_init_queues(sc);
/*
* Always get the chip out of the reset state, but only panic if not
* attaching. If attaching and there is an error, that is handled by
* the OS.
*/
error = mps_transition_operational(sc);
if (error != 0) {
if (attaching) {
mps_printf(sc, "%s failed to transition to operational "
"with error %d\n", __func__, error);
mps_free(sc);
return (error);
} else {
panic("%s failed to transition to operational with "
"error %d\n", __func__, error);
}
}
/*
* Finish the queue initialization.
* These are set here instead of in mps_init_queues() because the
* IOC resets these values during the state transition in
* mps_transition_operational(). The free index is set to 1
* because the corresponding index in the IOC is set to 0, and the
* IOC treats the queues as full if both are set to the same value.
* Hence the reason that the queue can't hold all of the possible
* replies.
*/
sc->replypostindex = 0;
mps_regwrite(sc, MPI2_REPLY_FREE_HOST_INDEX_OFFSET, sc->replyfreeindex);
mps_regwrite(sc, MPI2_REPLY_POST_HOST_INDEX_OFFSET, 0);
/*
* Attach the subsystems so they can prepare their event masks.
*/
/* XXX Should be dynamic so that IM/IR and user modules can attach */
if (attaching) {
if (((error = mps_attach_log(sc)) != 0) ||
((error = mps_attach_sas(sc)) != 0) ||
((error = mps_attach_user(sc)) != 0)) {
mps_printf(sc, "%s failed to attach all subsystems: "
"error %d\n", __func__, error);
mps_free(sc);
return (error);
}
if ((error = mps_pci_setup_interrupts(sc)) != 0) {
mps_printf(sc, "%s failed to setup interrupts\n",
__func__);
mps_free(sc);
return (error);
}
}
/*
* Set flag if this is a WD controller. This shouldn't ever change, but
* reset it after a Diag Reset, just in case.
*/
sc->WD_available = FALSE;
if (pci_get_device(sc->mps_dev) == MPI2_MFGPAGE_DEVID_SSS6200)
sc->WD_available = TRUE;
return (error);
}
/*
* This is called if memory is being free (during detach for example) and when
* buffers need to be reallocated due to a Diag Reset.
*/
static void
mps_iocfacts_free(struct mps_softc *sc)
{
struct mps_command *cm;
int i;
mps_dprint(sc, MPS_TRACE, "%s\n", __func__);
if (sc->post_busaddr != 0)
bus_dmamap_unload(sc->queues_dmat, sc->queues_map);
if (sc->post_queue != NULL)
bus_dmamem_free(sc->queues_dmat, sc->post_queue,
sc->queues_map);
if (sc->queues_dmat != NULL)
bus_dma_tag_destroy(sc->queues_dmat);
if (sc->chain_busaddr != 0)
bus_dmamap_unload(sc->chain_dmat, sc->chain_map);
if (sc->chain_frames != NULL)
bus_dmamem_free(sc->chain_dmat, sc->chain_frames,
sc->chain_map);
if (sc->chain_dmat != NULL)
bus_dma_tag_destroy(sc->chain_dmat);
if (sc->sense_busaddr != 0)
bus_dmamap_unload(sc->sense_dmat, sc->sense_map);
if (sc->sense_frames != NULL)
bus_dmamem_free(sc->sense_dmat, sc->sense_frames,
sc->sense_map);
if (sc->sense_dmat != NULL)
bus_dma_tag_destroy(sc->sense_dmat);
if (sc->reply_busaddr != 0)
bus_dmamap_unload(sc->reply_dmat, sc->reply_map);
if (sc->reply_frames != NULL)
bus_dmamem_free(sc->reply_dmat, sc->reply_frames,
sc->reply_map);
if (sc->reply_dmat != NULL)
bus_dma_tag_destroy(sc->reply_dmat);
if (sc->req_busaddr != 0)
bus_dmamap_unload(sc->req_dmat, sc->req_map);
if (sc->req_frames != NULL)
bus_dmamem_free(sc->req_dmat, sc->req_frames, sc->req_map);
if (sc->req_dmat != NULL)
bus_dma_tag_destroy(sc->req_dmat);
if (sc->chains != NULL)
free(sc->chains, M_MPT2);
if (sc->commands != NULL) {
for (i = 1; i < sc->num_reqs; i++) {
cm = &sc->commands[i];
bus_dmamap_destroy(sc->buffer_dmat, cm->cm_dmamap);
}
free(sc->commands, M_MPT2);
}
if (sc->buffer_dmat != NULL)
bus_dma_tag_destroy(sc->buffer_dmat);
}
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
/*
* The terms diag reset and hard reset are used interchangeably in the MPI
* docs to mean resetting the controller chip. In this code diag reset
* cleans everything up, and the hard reset function just sends the reset
* sequence to the chip. This should probably be refactored so that every
* subsystem gets a reset notification of some sort, and can clean up
* appropriately.
*/
int
mps_reinit(struct mps_softc *sc)
{
int error;
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
MPS_FUNCTRACE(sc);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
mtx_assert(&sc->mps_mtx, MA_OWNED);
if (sc->mps_flags & MPS_FLAGS_DIAGRESET) {
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_INIT, "%s reset already in progress\n",
__func__);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
return 0;
}
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_INFO, "Reinitializing controller,\n");
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
/* make sure the completion callbacks can recognize they're getting
* a NULL cm_reply due to a reset.
*/
sc->mps_flags |= MPS_FLAGS_DIAGRESET;
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
/*
* Mask interrupts here.
*/
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_INIT, "%s mask interrupts\n", __func__);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
mps_mask_intr(sc);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
error = mps_diag_reset(sc, CAN_SLEEP);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
if (error != 0) {
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
/* XXXSL No need to panic here */
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
panic("%s hard reset failed with error %d\n",
__func__, error);
}
/* Restore the PCI state, including the MSI-X registers */
mps_pci_restore(sc);
/* Give the I/O subsystem special priority to get itself prepared */
mpssas_handle_reinit(sc);
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
/*
* Get IOC Facts and allocate all structures based on this information.
* The attach function will also call mps_iocfacts_allocate at startup.
* If relevant values have changed in IOC Facts, this function will free
* all of the memory based on IOC Facts and reallocate that memory.
*/
if ((error = mps_iocfacts_allocate(sc, FALSE)) != 0) {
panic("%s IOC Facts based allocation failed with error %d\n",
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
__func__, error);
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
}
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
/*
* Mapping structures will be re-allocated after getting IOC Page8, so
* free these structures here.
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
*/
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
mps_mapping_exit(sc);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
/*
* The static page function currently read is IOC Page8. Others can be
* added in future. It's possible that the values in IOC Page8 have
* changed after a Diag Reset due to user modification, so always read
* these. Interrupts are masked, so unmask them before getting config
* pages.
*/
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
mps_unmask_intr(sc);
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
sc->mps_flags &= ~MPS_FLAGS_DIAGRESET;
mps_base_static_config_pages(sc);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
/*
* Some mapping info is based in IOC Page8 data, so re-initialize the
* mapping tables.
*/
mps_mapping_initialize(sc);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
/*
* Restart will reload the event masks clobbered by the reset, and
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
* then enable the port.
*/
mps_reregister_events(sc);
/* the end of discovery will release the simq, so we're done. */
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_INFO, "%s finished sc %p post %u free %u\n",
__func__, sc, sc->replypostindex, sc->replyfreeindex);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
return 0;
}
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
/* Wait for the chip to ACK a word that we've put into its FIFO
* Wait for <timeout> seconds. In single loop wait for busy loop
* for 500 microseconds.
* Total is [ 0.5 * (2000 * <timeout>) ] in miliseconds.
* */
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
static int
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
mps_wait_db_ack(struct mps_softc *sc, int timeout, int sleep_flag)
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
{
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
u32 cntdn, count;
u32 int_status;
u32 doorbell;
count = 0;
cntdn = (sleep_flag == CAN_SLEEP) ? 1000*timeout : 2000*timeout;
do {
int_status = mps_regread(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET);
if (!(int_status & MPI2_HIS_SYS2IOC_DB_STATUS)) {
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_INIT,
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
"%s: successfull count(%d), timeout(%d)\n",
__func__, count, timeout);
return 0;
} else if (int_status & MPI2_HIS_IOC2SYS_DB_STATUS) {
doorbell = mps_regread(sc, MPI2_DOORBELL_OFFSET);
if ((doorbell & MPI2_IOC_STATE_MASK) ==
MPI2_IOC_STATE_FAULT) {
mps_dprint(sc, MPS_FAULT,
"fault_state(0x%04x)!\n", doorbell);
return (EFAULT);
}
} else if (int_status == 0xFFFFFFFF)
goto out;
/* If it can sleep, sleep for 1 milisecond, else busy loop for
* 0.5 milisecond */
if (mtx_owned(&sc->mps_mtx) && sleep_flag == CAN_SLEEP)
msleep(&sc->msleep_fake_chan, &sc->mps_mtx, 0,
"mpsdba", hz/1000);
else if (sleep_flag == CAN_SLEEP)
pause("mpsdba", hz/1000);
else
DELAY(500);
count++;
} while (--cntdn);
out:
mps_dprint(sc, MPS_FAULT, "%s: failed due to timeout count(%d), "
"int_status(%x)!\n", __func__, count, int_status);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
return (ETIMEDOUT);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
}
/* Wait for the chip to signal that the next word in its FIFO can be fetched */
static int
mps_wait_db_int(struct mps_softc *sc)
{
int retry;
for (retry = 0; retry < MPS_DB_MAX_WAIT; retry++) {
if ((mps_regread(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET) &
MPI2_HIS_IOC2SYS_DB_STATUS) != 0)
return (0);
DELAY(2000);
}
return (ETIMEDOUT);
}
/* Step through the synchronous command state machine, i.e. "Doorbell mode" */
static int
mps_request_sync(struct mps_softc *sc, void *req, MPI2_DEFAULT_REPLY *reply,
int req_sz, int reply_sz, int timeout)
{
uint32_t *data32;
uint16_t *data16;
int i, count, ioc_sz, residual;
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
int sleep_flags = CAN_SLEEP;
Replace the TDP_NOSLEEPING flag with a counter so that the THREAD_NO_SLEEPING() and THREAD_SLEEPING_OK() macros can nest. Reviewed by: attilio
2013-03-01 22:03:31 +00:00
if (curthread->td_no_sleeping != 0)
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
sleep_flags = NO_SLEEP;
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
/* Step 1 */
mps_regwrite(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET, 0x0);
/* Step 2 */
if (mps_regread(sc, MPI2_DOORBELL_OFFSET) & MPI2_DOORBELL_USED)
return (EBUSY);
/* Step 3
* Announce that a message is coming through the doorbell. Messages
* are pushed at 32bit words, so round up if needed.
*/
count = (req_sz + 3) / 4;
mps_regwrite(sc, MPI2_DOORBELL_OFFSET,
(MPI2_FUNCTION_HANDSHAKE << MPI2_DOORBELL_FUNCTION_SHIFT) |
(count << MPI2_DOORBELL_ADD_DWORDS_SHIFT));
/* Step 4 */
if (mps_wait_db_int(sc) ||
(mps_regread(sc, MPI2_DOORBELL_OFFSET) & MPI2_DOORBELL_USED) == 0) {
mps_dprint(sc, MPS_FAULT, "Doorbell failed to activate\n");
return (ENXIO);
}
mps_regwrite(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET, 0x0);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
if (mps_wait_db_ack(sc, 5, sleep_flags) != 0) {
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
mps_dprint(sc, MPS_FAULT, "Doorbell handshake failed\n");
return (ENXIO);
}
/* Step 5 */
/* Clock out the message data synchronously in 32-bit dwords*/
data32 = (uint32_t *)req;
for (i = 0; i < count; i++) {
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
mps_regwrite(sc, MPI2_DOORBELL_OFFSET, htole32(data32[i]));
if (mps_wait_db_ack(sc, 5, sleep_flags) != 0) {
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
mps_dprint(sc, MPS_FAULT,
"Timeout while writing doorbell\n");
return (ENXIO);
}
}
/* Step 6 */
/* Clock in the reply in 16-bit words. The total length of the
* message is always in the 4th byte, so clock out the first 2 words
* manually, then loop the rest.
*/
data16 = (uint16_t *)reply;
if (mps_wait_db_int(sc) != 0) {
mps_dprint(sc, MPS_FAULT, "Timeout reading doorbell 0\n");
return (ENXIO);
}
data16[0] =
mps_regread(sc, MPI2_DOORBELL_OFFSET) & MPI2_DOORBELL_DATA_MASK;
mps_regwrite(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET, 0x0);
if (mps_wait_db_int(sc) != 0) {
mps_dprint(sc, MPS_FAULT, "Timeout reading doorbell 1\n");
return (ENXIO);
}
data16[1] =
mps_regread(sc, MPI2_DOORBELL_OFFSET) & MPI2_DOORBELL_DATA_MASK;
mps_regwrite(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET, 0x0);
/* Number of 32bit words in the message */
ioc_sz = reply->MsgLength;
/*
* Figure out how many 16bit words to clock in without overrunning.
* The precision loss with dividing reply_sz can safely be
* ignored because the messages can only be multiples of 32bits.
*/
residual = 0;
count = MIN((reply_sz / 4), ioc_sz) * 2;
if (count < ioc_sz * 2) {
residual = ioc_sz * 2 - count;
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_ERROR, "Driver error, throwing away %d "
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
"residual message words\n", residual);
}
for (i = 2; i < count; i++) {
if (mps_wait_db_int(sc) != 0) {
mps_dprint(sc, MPS_FAULT,
"Timeout reading doorbell %d\n", i);
return (ENXIO);
}
data16[i] = mps_regread(sc, MPI2_DOORBELL_OFFSET) &
MPI2_DOORBELL_DATA_MASK;
mps_regwrite(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET, 0x0);
}
/*
* Pull out residual words that won't fit into the provided buffer.
* This keeps the chip from hanging due to a driver programming
* error.
*/
while (residual--) {
if (mps_wait_db_int(sc) != 0) {
mps_dprint(sc, MPS_FAULT,
"Timeout reading doorbell\n");
return (ENXIO);
}
(void)mps_regread(sc, MPI2_DOORBELL_OFFSET);
mps_regwrite(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET, 0x0);
}
/* Step 7 */
if (mps_wait_db_int(sc) != 0) {
mps_dprint(sc, MPS_FAULT, "Timeout waiting to exit doorbell\n");
return (ENXIO);
}
if (mps_regread(sc, MPI2_DOORBELL_OFFSET) & MPI2_DOORBELL_USED)
mps_dprint(sc, MPS_FAULT, "Warning, doorbell still active\n");
mps_regwrite(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET, 0x0);
return (0);
}
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
static void
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
mps_enqueue_request(struct mps_softc *sc, struct mps_command *cm)
{
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
reply_descriptor rd;
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
MPS_FUNCTRACE(sc);
mps_dprint(sc, MPS_TRACE, "SMID %u cm %p ccb %p\n",
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
cm->cm_desc.Default.SMID, cm, cm->cm_ccb);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Bring in a number of mps(4) driver fixes from LSI: 1. Fixed timeout specification for the msleep in mps_wait_command(). Added 30 second timeout for mps_wait_command() calls in mps_user.c. 2. Make sure we call mps_detach_user() from the kldunload path. 3. Raid Hotplug behavior change. The driver now removes a volume when it goes to a failed state, so we also need to add volume back to the OS when it goes to opitimal/degraded/online from failed/missing. Handle raid volume add and remove from the IR_Volume event. 4. Added some more debugging information. 5. Replace xpt_async(AC_LOST_DEVICE, path, NULL) with mpssas_rescan_target(). This is to work around a panic in CAM that shows up when adding a drive with a rescan and removing another device from the driver thread with an AC_LOST_DEVICE async notification. This problem was encountered in testing with the LSI sas2ircu utility, which was used to create a RAID volume from physical disks. The driver has to create the RAID volume target and remove the physical disk targets, and triggered a panic in the process. The CAM issue needs to be fully diagnosed and fixed, but this works around the issue for now. 6. Fix some memory initialization issues in mps_free_command(). 7. Resolve the "devq freeze forever" issue. This was caused by the internal read capacity command issued in the non-head version of the driver. When the command completed with an error, the driver wasn't unfreezing thd device queue. The version in head uses the CAM infrastructure for getting the read capacity information, and therefore doesn't have the same issue. 8. Bump the version to 13.00.00.00-fbsd. (this is very close to LSI's internal stable driver 13.00.00.00) Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> MFC after: 3 days
2012-02-09 00:16:12 +00:00
if (sc->mps_flags & MPS_FLAGS_ATTACH_DONE && !(sc->mps_flags & MPS_FLAGS_SHUTDOWN))
Fix several issues with the mps(4) driver. When the driver ran out of DMA chaining buffers, it kept the timeout for the I/O, and I/O would stall. The driver was not freezing the device queue on errors. mps.c: Pull command completion logic into a separate function, and call the callback/wakeup for commands that are never sent due to lack of chain buffers. Add a number of extra diagnostic sysctl variables. Handle pre-hardware errors for configuration I/O. This doesn't panic the system, but it will fail the configuration I/O and there is no retry mechanism. So the device probe will not succeed. This should be a very uncommon situation, however. mps_sas.c: Freeze the SIM queue when we run out of chain buffers, and unfreeze it when more commands complete. Freeze the device queue when errors occur, so that CAM can insure proper command ordering. Report pre-hardware errors for task management commands. In general, that shouldn't be possible because task management commands don't have S/G lists, and that is currently the only error path before we get to the hardware. Handle pre-hardware errors (like out of chain elements) for SMP requests. That shouldn't happen either, since we should have enough space for two S/G elements in the standard request. For commands that end with MPI2_IOCSTATUS_SCSI_IOC_TERMINATED and MPI2_IOCSTATUS_SCSI_EXT_TERMINATED, return them with CAM_REQUEUE_REQ to retry them unconditionally. These seem to be related to back end, transport related problems that are hopefully transient. We don't want to go through the retry count for something that is not a permanent error. Keep track of the number of outstanding I/Os. mpsvar.h: Track the number of free chain elements. Add variables for the number of outstanding I/Os, and I/O high water mark. Add variables to track the number of free chain buffers and the chain low water mark, as well as the number of chain allocation failures. Add I/O state flags and an attach done flag. MFC after: 3 days
2011-02-18 17:06:06 +00:00
mtx_assert(&sc->mps_mtx, MA_OWNED);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
if (++sc->io_cmds_active > sc->io_cmds_highwater)
sc->io_cmds_highwater++;
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
rd.u.low = cm->cm_desc.Words.Low;
rd.u.high = cm->cm_desc.Words.High;
rd.word = htole64(rd.word);
/* TODO-We may need to make below regwrite atomic */
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
mps_regwrite(sc, MPI2_REQUEST_DESCRIPTOR_POST_LOW_OFFSET,
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
rd.u.low);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
mps_regwrite(sc, MPI2_REQUEST_DESCRIPTOR_POST_HIGH_OFFSET,
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
rd.u.high);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
}
/*
* Just the FACTS, ma'am.
*/
static int
mps_get_iocfacts(struct mps_softc *sc, MPI2_IOC_FACTS_REPLY *facts)
{
MPI2_DEFAULT_REPLY *reply;
MPI2_IOC_FACTS_REQUEST request;
int error, req_sz, reply_sz;
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
MPS_FUNCTRACE(sc);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
req_sz = sizeof(MPI2_IOC_FACTS_REQUEST);
reply_sz = sizeof(MPI2_IOC_FACTS_REPLY);
reply = (MPI2_DEFAULT_REPLY *)facts;
bzero(&request, req_sz);
request.Function = MPI2_FUNCTION_IOC_FACTS;
error = mps_request_sync(sc, &request, reply, req_sz, reply_sz, 5);
return (error);
}
static int
mps_send_iocinit(struct mps_softc *sc)
{
MPI2_IOC_INIT_REQUEST init;
MPI2_DEFAULT_REPLY reply;
int req_sz, reply_sz, error;
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
struct timeval now;
uint64_t time_in_msec;
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
MPS_FUNCTRACE(sc);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
req_sz = sizeof(MPI2_IOC_INIT_REQUEST);
reply_sz = sizeof(MPI2_IOC_INIT_REPLY);
bzero(&init, req_sz);
bzero(&reply, reply_sz);
/*
* Fill in the init block. Note that most addresses are
* deliberately in the lower 32bits of memory. This is a micro-
* optimzation for PCI/PCIX, though it's not clear if it helps PCIe.
*/
init.Function = MPI2_FUNCTION_IOC_INIT;
init.WhoInit = MPI2_WHOINIT_HOST_DRIVER;
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
init.MsgVersion = htole16(MPI2_VERSION);
init.HeaderVersion = htole16(MPI2_HEADER_VERSION);
init.SystemRequestFrameSize = htole16(sc->facts->IOCRequestFrameSize);
init.ReplyDescriptorPostQueueDepth = htole16(sc->pqdepth);
init.ReplyFreeQueueDepth = htole16(sc->fqdepth);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
init.SenseBufferAddressHigh = 0;
init.SystemReplyAddressHigh = 0;
init.SystemRequestFrameBaseAddress.High = 0;
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
init.SystemRequestFrameBaseAddress.Low = htole32((uint32_t)sc->req_busaddr);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
init.ReplyDescriptorPostQueueAddress.High = 0;
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
init.ReplyDescriptorPostQueueAddress.Low = htole32((uint32_t)sc->post_busaddr);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
init.ReplyFreeQueueAddress.High = 0;
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
init.ReplyFreeQueueAddress.Low = htole32((uint32_t)sc->free_busaddr);
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
getmicrotime(&now);
time_in_msec = (now.tv_sec * 1000 + now.tv_usec/1000);
init.TimeStamp.High = htole32((time_in_msec >> 32) & 0xFFFFFFFF);
init.TimeStamp.Low = htole32(time_in_msec & 0xFFFFFFFF);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
error = mps_request_sync(sc, &init, &reply, req_sz, reply_sz, 5);
if ((reply.IOCStatus & MPI2_IOCSTATUS_MASK) != MPI2_IOCSTATUS_SUCCESS)
error = ENXIO;
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_INIT, "IOCInit status= 0x%x\n", reply.IOCStatus);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
return (error);
}
void
mps_memaddr_cb(void *arg, bus_dma_segment_t *segs, int nsegs, int error)
{
bus_addr_t *addr;
addr = arg;
*addr = segs[0].ds_addr;
}
static int
mps_alloc_queues(struct mps_softc *sc)
{
bus_addr_t queues_busaddr;
uint8_t *queues;
int qsize, fqsize, pqsize;
/*
* The reply free queue contains 4 byte entries in multiples of 16 and
* aligned on a 16 byte boundary. There must always be an unused entry.
* This queue supplies fresh reply frames for the firmware to use.
*
* The reply descriptor post queue contains 8 byte entries in
* multiples of 16 and aligned on a 16 byte boundary. This queue
* contains filled-in reply frames sent from the firmware to the host.
*
* These two queues are allocated together for simplicity.
*/
sc->fqdepth = roundup2((sc->num_replies + 1), 16);
sc->pqdepth = roundup2((sc->num_replies + 1), 16);
fqsize= sc->fqdepth * 4;
pqsize = sc->pqdepth * 8;
qsize = fqsize + pqsize;
if (bus_dma_tag_create( sc->mps_parent_dmat, /* parent */
16, 0, /* algnmnt, boundary */
BUS_SPACE_MAXADDR_32BIT,/* lowaddr */
BUS_SPACE_MAXADDR, /* highaddr */
NULL, NULL, /* filter, filterarg */
qsize, /* maxsize */
1, /* nsegments */
qsize, /* maxsegsize */
0, /* flags */
NULL, NULL, /* lockfunc, lockarg */
&sc->queues_dmat)) {
device_printf(sc->mps_dev, "Cannot allocate queues DMA tag\n");
return (ENOMEM);
}
if (bus_dmamem_alloc(sc->queues_dmat, (void **)&queues, BUS_DMA_NOWAIT,
&sc->queues_map)) {
device_printf(sc->mps_dev, "Cannot allocate queues memory\n");
return (ENOMEM);
}
bzero(queues, qsize);
bus_dmamap_load(sc->queues_dmat, sc->queues_map, queues, qsize,
mps_memaddr_cb, &queues_busaddr, 0);
sc->free_queue = (uint32_t *)queues;
sc->free_busaddr = queues_busaddr;
sc->post_queue = (MPI2_REPLY_DESCRIPTORS_UNION *)(queues + fqsize);
sc->post_busaddr = queues_busaddr + fqsize;
return (0);
}
static int
mps_alloc_replies(struct mps_softc *sc)
{
Fix an event handling bug with the mps(4) driver. This bug manifested itself after repeated device arrivals and departures. The root of the problem was that the last entry in the reply array wasn't initialized/allocated. So every time we got around to that event, we had a bogus address. There were a couple more problems with the code that are also fixed: - The reply mechanism was being treated as sequential (indexed by sc->replycurindex) even though the spec says that the driver should use the ReplyFrameAddress field of the post queue descriptor to figure out where the reply is. There is no guarantee that the reply descriptors will be used in sequential order. - The second word of the reply post queue descriptor wasn't being checked in mps_intr_locked() to make sure that it wasn't 0xffffffff. So the driver could potentially come across a partially DMAed descriptor. - The number of replies allocated was one less than the actual size of the queue. Instead, it was the size of the number of replies that can be used at one time. (Which is one less than the size of the queue.) mps.c: When initializing the entries in the reply free queue, make sure we initialize the full number that we tell the chip we have (sc->fqdepth), not the number that can be used at any one time (sc->num_replies). When allocating replies, make sure we allocate the number of replies that we've told the chip exist, not just the number that can be used simultaneously. Use the ReplyFrameAddress field of the post queue descriptor to figure out which reply is being referenced. This is what the spec says to do, and the spec doesn't guarantee that the replies will be used in order. Put a check in to verify that the reply address passed back from the card is valid. (Panic if it isn't, we'll panic when we try to deference the reply pointer in any case.) In mps_intr_locked(), verify that the second word of the post queue descriptor is not 0xffffffff in addition to verifying that the unused flag is not set, so we can make sure we didn't get a partially DMAed descriptor. Remove references to sc->replycurindex, it isn't needed now. mpsvar.h: Remove replycurindex from the softc, it isn't needed now. Reviewed by: scottl
2010-12-10 21:45:10 +00:00
int rsize, num_replies;
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Fix an event handling bug with the mps(4) driver. This bug manifested itself after repeated device arrivals and departures. The root of the problem was that the last entry in the reply array wasn't initialized/allocated. So every time we got around to that event, we had a bogus address. There were a couple more problems with the code that are also fixed: - The reply mechanism was being treated as sequential (indexed by sc->replycurindex) even though the spec says that the driver should use the ReplyFrameAddress field of the post queue descriptor to figure out where the reply is. There is no guarantee that the reply descriptors will be used in sequential order. - The second word of the reply post queue descriptor wasn't being checked in mps_intr_locked() to make sure that it wasn't 0xffffffff. So the driver could potentially come across a partially DMAed descriptor. - The number of replies allocated was one less than the actual size of the queue. Instead, it was the size of the number of replies that can be used at one time. (Which is one less than the size of the queue.) mps.c: When initializing the entries in the reply free queue, make sure we initialize the full number that we tell the chip we have (sc->fqdepth), not the number that can be used at any one time (sc->num_replies). When allocating replies, make sure we allocate the number of replies that we've told the chip exist, not just the number that can be used simultaneously. Use the ReplyFrameAddress field of the post queue descriptor to figure out which reply is being referenced. This is what the spec says to do, and the spec doesn't guarantee that the replies will be used in order. Put a check in to verify that the reply address passed back from the card is valid. (Panic if it isn't, we'll panic when we try to deference the reply pointer in any case.) In mps_intr_locked(), verify that the second word of the post queue descriptor is not 0xffffffff in addition to verifying that the unused flag is not set, so we can make sure we didn't get a partially DMAed descriptor. Remove references to sc->replycurindex, it isn't needed now. mpsvar.h: Remove replycurindex from the softc, it isn't needed now. Reviewed by: scottl
2010-12-10 21:45:10 +00:00
/*
* sc->num_replies should be one less than sc->fqdepth. We need to
* allocate space for sc->fqdepth replies, but only sc->num_replies
* replies can be used at once.
*/
num_replies = max(sc->fqdepth, sc->num_replies);
rsize = sc->facts->ReplyFrameSize * num_replies * 4;
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
if (bus_dma_tag_create( sc->mps_parent_dmat, /* parent */
4, 0, /* algnmnt, boundary */
BUS_SPACE_MAXADDR_32BIT,/* lowaddr */
BUS_SPACE_MAXADDR, /* highaddr */
NULL, NULL, /* filter, filterarg */
rsize, /* maxsize */
1, /* nsegments */
rsize, /* maxsegsize */
0, /* flags */
NULL, NULL, /* lockfunc, lockarg */
&sc->reply_dmat)) {
device_printf(sc->mps_dev, "Cannot allocate replies DMA tag\n");
return (ENOMEM);
}
if (bus_dmamem_alloc(sc->reply_dmat, (void **)&sc->reply_frames,
BUS_DMA_NOWAIT, &sc->reply_map)) {
device_printf(sc->mps_dev, "Cannot allocate replies memory\n");
return (ENOMEM);
}
bzero(sc->reply_frames, rsize);
bus_dmamap_load(sc->reply_dmat, sc->reply_map, sc->reply_frames, rsize,
mps_memaddr_cb, &sc->reply_busaddr, 0);
return (0);
}
static int
mps_alloc_requests(struct mps_softc *sc)
{
struct mps_command *cm;
struct mps_chain *chain;
int i, rsize, nsegs;
rsize = sc->facts->IOCRequestFrameSize * sc->num_reqs * 4;
if (bus_dma_tag_create( sc->mps_parent_dmat, /* parent */
16, 0, /* algnmnt, boundary */
BUS_SPACE_MAXADDR_32BIT,/* lowaddr */
BUS_SPACE_MAXADDR, /* highaddr */
NULL, NULL, /* filter, filterarg */
rsize, /* maxsize */
1, /* nsegments */
rsize, /* maxsegsize */
0, /* flags */
NULL, NULL, /* lockfunc, lockarg */
&sc->req_dmat)) {
device_printf(sc->mps_dev, "Cannot allocate request DMA tag\n");
return (ENOMEM);
}
if (bus_dmamem_alloc(sc->req_dmat, (void **)&sc->req_frames,
BUS_DMA_NOWAIT, &sc->req_map)) {
device_printf(sc->mps_dev, "Cannot allocate request memory\n");
return (ENOMEM);
}
bzero(sc->req_frames, rsize);
bus_dmamap_load(sc->req_dmat, sc->req_map, sc->req_frames, rsize,
mps_memaddr_cb, &sc->req_busaddr, 0);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
rsize = sc->facts->IOCRequestFrameSize * sc->max_chains * 4;
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
if (bus_dma_tag_create( sc->mps_parent_dmat, /* parent */
16, 0, /* algnmnt, boundary */
BUS_SPACE_MAXADDR_32BIT,/* lowaddr */
BUS_SPACE_MAXADDR, /* highaddr */
NULL, NULL, /* filter, filterarg */
rsize, /* maxsize */
1, /* nsegments */
rsize, /* maxsegsize */
0, /* flags */
NULL, NULL, /* lockfunc, lockarg */
&sc->chain_dmat)) {
device_printf(sc->mps_dev, "Cannot allocate chain DMA tag\n");
return (ENOMEM);
}
if (bus_dmamem_alloc(sc->chain_dmat, (void **)&sc->chain_frames,
BUS_DMA_NOWAIT, &sc->chain_map)) {
device_printf(sc->mps_dev, "Cannot allocate chain memory\n");
return (ENOMEM);
}
bzero(sc->chain_frames, rsize);
bus_dmamap_load(sc->chain_dmat, sc->chain_map, sc->chain_frames, rsize,
mps_memaddr_cb, &sc->chain_busaddr, 0);
rsize = MPS_SENSE_LEN * sc->num_reqs;
if (bus_dma_tag_create( sc->mps_parent_dmat, /* parent */
1, 0, /* algnmnt, boundary */
BUS_SPACE_MAXADDR_32BIT,/* lowaddr */
BUS_SPACE_MAXADDR, /* highaddr */
NULL, NULL, /* filter, filterarg */
rsize, /* maxsize */
1, /* nsegments */
rsize, /* maxsegsize */
0, /* flags */
NULL, NULL, /* lockfunc, lockarg */
&sc->sense_dmat)) {
device_printf(sc->mps_dev, "Cannot allocate sense DMA tag\n");
return (ENOMEM);
}
if (bus_dmamem_alloc(sc->sense_dmat, (void **)&sc->sense_frames,
BUS_DMA_NOWAIT, &sc->sense_map)) {
device_printf(sc->mps_dev, "Cannot allocate sense memory\n");
return (ENOMEM);
}
bzero(sc->sense_frames, rsize);
bus_dmamap_load(sc->sense_dmat, sc->sense_map, sc->sense_frames, rsize,
mps_memaddr_cb, &sc->sense_busaddr, 0);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
sc->chains = malloc(sizeof(struct mps_chain) * sc->max_chains, M_MPT2,
M_WAITOK | M_ZERO);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
if(!sc->chains) {
device_printf(sc->mps_dev,
"Cannot allocate chains memory %s %d\n",
__func__, __LINE__);
return (ENOMEM);
}
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
for (i = 0; i < sc->max_chains; i++) {
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
chain = &sc->chains[i];
chain->chain = (MPI2_SGE_IO_UNION *)(sc->chain_frames +
i * sc->facts->IOCRequestFrameSize * 4);
chain->chain_busaddr = sc->chain_busaddr +
i * sc->facts->IOCRequestFrameSize * 4;
mps_free_chain(sc, chain);
Fix several issues with the mps(4) driver. When the driver ran out of DMA chaining buffers, it kept the timeout for the I/O, and I/O would stall. The driver was not freezing the device queue on errors. mps.c: Pull command completion logic into a separate function, and call the callback/wakeup for commands that are never sent due to lack of chain buffers. Add a number of extra diagnostic sysctl variables. Handle pre-hardware errors for configuration I/O. This doesn't panic the system, but it will fail the configuration I/O and there is no retry mechanism. So the device probe will not succeed. This should be a very uncommon situation, however. mps_sas.c: Freeze the SIM queue when we run out of chain buffers, and unfreeze it when more commands complete. Freeze the device queue when errors occur, so that CAM can insure proper command ordering. Report pre-hardware errors for task management commands. In general, that shouldn't be possible because task management commands don't have S/G lists, and that is currently the only error path before we get to the hardware. Handle pre-hardware errors (like out of chain elements) for SMP requests. That shouldn't happen either, since we should have enough space for two S/G elements in the standard request. For commands that end with MPI2_IOCSTATUS_SCSI_IOC_TERMINATED and MPI2_IOCSTATUS_SCSI_EXT_TERMINATED, return them with CAM_REQUEUE_REQ to retry them unconditionally. These seem to be related to back end, transport related problems that are hopefully transient. We don't want to go through the retry count for something that is not a permanent error. Keep track of the number of outstanding I/Os. mpsvar.h: Track the number of free chain elements. Add variables for the number of outstanding I/Os, and I/O high water mark. Add variables to track the number of free chain buffers and the chain low water mark, as well as the number of chain allocation failures. Add I/O state flags and an attach done flag. MFC after: 3 days
2011-02-18 17:06:06 +00:00
sc->chain_free_lowwater++;
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
}
/* XXX Need to pick a more precise value */
nsegs = (MAXPHYS / PAGE_SIZE) + 1;
if (bus_dma_tag_create( sc->mps_parent_dmat, /* parent */
1, 0, /* algnmnt, boundary */
BUS_SPACE_MAXADDR, /* lowaddr */
BUS_SPACE_MAXADDR, /* highaddr */
NULL, NULL, /* filter, filterarg */
BUS_SPACE_MAXSIZE_32BIT,/* maxsize */
nsegs, /* nsegments */
Several fixes to allow firmware/BIOS flash access from user-level: - remove special handling of zero length transfers in mpi_pre_fw_upload(); - add missing MPS_CM_FLAGS_DATAIN flag in mpi_pre_fw_upload(); - move mps_user_setup_request() call into proper place; - increase user command timeout from 30 to 60 seconds; - avoid NULL dereference panic in case of firmware crash. Set max DMA segment size to 24bit, as MPI SGE supports it. Use mps_add_dmaseg() to add empty SGE instead of custom code. Tune endianness safety. Reviewed by: Desai, Kashyap <Kashyap.Desai@lsi.com> Sponsored by: iXsystems, Inc.
2012-08-01 17:31:31 +00:00
BUS_SPACE_MAXSIZE_24BIT,/* maxsegsize */
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
BUS_DMA_ALLOCNOW, /* flags */
busdma_lock_mutex, /* lockfunc */
&sc->mps_mtx, /* lockarg */
&sc->buffer_dmat)) {
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
device_printf(sc->mps_dev, "Cannot allocate buffer DMA tag\n");
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
return (ENOMEM);
}
/*
* SMID 0 cannot be used as a free command per the firmware spec.
* Just drop that command instead of risking accounting bugs.
*/
sc->commands = malloc(sizeof(struct mps_command) * sc->num_reqs,
M_MPT2, M_WAITOK | M_ZERO);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
if(!sc->commands) {
device_printf(sc->mps_dev, "Cannot allocate memory %s %d\n",
__func__, __LINE__);
return (ENOMEM);
}
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
for (i = 1; i < sc->num_reqs; i++) {
cm = &sc->commands[i];
cm->cm_req = sc->req_frames +
i * sc->facts->IOCRequestFrameSize * 4;
cm->cm_req_busaddr = sc->req_busaddr +
i * sc->facts->IOCRequestFrameSize * 4;
cm->cm_sense = &sc->sense_frames[i];
cm->cm_sense_busaddr = sc->sense_busaddr + i * MPS_SENSE_LEN;
cm->cm_desc.Default.SMID = i;
cm->cm_sc = sc;
TAILQ_INIT(&cm->cm_chain_list);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
callout_init_mtx(&cm->cm_callout, &sc->mps_mtx, 0);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
/* XXX Is a failure here a critical problem? */
if (bus_dmamap_create(sc->buffer_dmat, 0, &cm->cm_dmamap) == 0)
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
if (i <= sc->facts->HighPriorityCredit)
mps_free_high_priority_command(sc, cm);
else
mps_free_command(sc, cm);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
else {
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
panic("failed to allocate command %d\n", i);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
sc->num_reqs = i;
break;
}
}
return (0);
}
static int
mps_init_queues(struct mps_softc *sc)
{
int i;
memset((uint8_t *)sc->post_queue, 0xff, sc->pqdepth * 8);
Fix an event handling bug with the mps(4) driver. This bug manifested itself after repeated device arrivals and departures. The root of the problem was that the last entry in the reply array wasn't initialized/allocated. So every time we got around to that event, we had a bogus address. There were a couple more problems with the code that are also fixed: - The reply mechanism was being treated as sequential (indexed by sc->replycurindex) even though the spec says that the driver should use the ReplyFrameAddress field of the post queue descriptor to figure out where the reply is. There is no guarantee that the reply descriptors will be used in sequential order. - The second word of the reply post queue descriptor wasn't being checked in mps_intr_locked() to make sure that it wasn't 0xffffffff. So the driver could potentially come across a partially DMAed descriptor. - The number of replies allocated was one less than the actual size of the queue. Instead, it was the size of the number of replies that can be used at one time. (Which is one less than the size of the queue.) mps.c: When initializing the entries in the reply free queue, make sure we initialize the full number that we tell the chip we have (sc->fqdepth), not the number that can be used at any one time (sc->num_replies). When allocating replies, make sure we allocate the number of replies that we've told the chip exist, not just the number that can be used simultaneously. Use the ReplyFrameAddress field of the post queue descriptor to figure out which reply is being referenced. This is what the spec says to do, and the spec doesn't guarantee that the replies will be used in order. Put a check in to verify that the reply address passed back from the card is valid. (Panic if it isn't, we'll panic when we try to deference the reply pointer in any case.) In mps_intr_locked(), verify that the second word of the post queue descriptor is not 0xffffffff in addition to verifying that the unused flag is not set, so we can make sure we didn't get a partially DMAed descriptor. Remove references to sc->replycurindex, it isn't needed now. mpsvar.h: Remove replycurindex from the softc, it isn't needed now. Reviewed by: scottl
2010-12-10 21:45:10 +00:00
/*
* According to the spec, we need to use one less reply than we
* have space for on the queue. So sc->num_replies (the number we
* use) should be less than sc->fqdepth (allocated size).
*/
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
if (sc->num_replies >= sc->fqdepth)
return (EINVAL);
Fix an event handling bug with the mps(4) driver. This bug manifested itself after repeated device arrivals and departures. The root of the problem was that the last entry in the reply array wasn't initialized/allocated. So every time we got around to that event, we had a bogus address. There were a couple more problems with the code that are also fixed: - The reply mechanism was being treated as sequential (indexed by sc->replycurindex) even though the spec says that the driver should use the ReplyFrameAddress field of the post queue descriptor to figure out where the reply is. There is no guarantee that the reply descriptors will be used in sequential order. - The second word of the reply post queue descriptor wasn't being checked in mps_intr_locked() to make sure that it wasn't 0xffffffff. So the driver could potentially come across a partially DMAed descriptor. - The number of replies allocated was one less than the actual size of the queue. Instead, it was the size of the number of replies that can be used at one time. (Which is one less than the size of the queue.) mps.c: When initializing the entries in the reply free queue, make sure we initialize the full number that we tell the chip we have (sc->fqdepth), not the number that can be used at any one time (sc->num_replies). When allocating replies, make sure we allocate the number of replies that we've told the chip exist, not just the number that can be used simultaneously. Use the ReplyFrameAddress field of the post queue descriptor to figure out which reply is being referenced. This is what the spec says to do, and the spec doesn't guarantee that the replies will be used in order. Put a check in to verify that the reply address passed back from the card is valid. (Panic if it isn't, we'll panic when we try to deference the reply pointer in any case.) In mps_intr_locked(), verify that the second word of the post queue descriptor is not 0xffffffff in addition to verifying that the unused flag is not set, so we can make sure we didn't get a partially DMAed descriptor. Remove references to sc->replycurindex, it isn't needed now. mpsvar.h: Remove replycurindex from the softc, it isn't needed now. Reviewed by: scottl
2010-12-10 21:45:10 +00:00
/*
* Initialize all of the free queue entries.
*/
for (i = 0; i < sc->fqdepth; i++)
sc->free_queue[i] = sc->reply_busaddr + (i * sc->facts->ReplyFrameSize * 4);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
sc->replyfreeindex = sc->num_replies;
return (0);
}
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
/* Get the driver parameter tunables. Lowest priority are the driver defaults.
* Next are the global settings, if they exist. Highest are the per-unit
* settings, if they exist.
*/
static void
mps_get_tunables(struct mps_softc *sc)
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
{
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
char tmpstr[80];
/* XXX default to some debugging for now */
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
sc->mps_debug = MPS_INFO|MPS_FAULT;
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
sc->disable_msix = 0;
sc->disable_msi = 0;
sc->max_chains = MPS_CHAIN_FRAMES;
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
/*
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
* Grab the global variables.
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
*/
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
TUNABLE_INT_FETCH("hw.mps.debug_level", &sc->mps_debug);
TUNABLE_INT_FETCH("hw.mps.disable_msix", &sc->disable_msix);
TUNABLE_INT_FETCH("hw.mps.disable_msi", &sc->disable_msi);
TUNABLE_INT_FETCH("hw.mps.max_chains", &sc->max_chains);
/* Grab the unit-instance variables */
snprintf(tmpstr, sizeof(tmpstr), "dev.mps.%d.debug_level",
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
device_get_unit(sc->mps_dev));
TUNABLE_INT_FETCH(tmpstr, &sc->mps_debug);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
snprintf(tmpstr, sizeof(tmpstr), "dev.mps.%d.disable_msix",
Turn on serialization of task management commands going down to the controller, but make it optional. After a problem report from Andrew Boyer, it looks like the LSI chip may have issues (the watchdog timer fired) if too many aborts are sent down to the chip at the same time. We know that task management commands are serialized, and although the manual doesn't say it, it may be a good idea to just send one at a time. But, since I'm not certain that this is necessary, add a tunable and sysctl variable (hw.mps.%d.allow_multiple_tm_cmds) to control the driver's behavior. mps.c: Add support for the sysctl and tunable, and add a comment about the possible return values to mps_map_command(). mps_sas.c: Run all task management commands through two new routines, mpssas_issue_tm_request() and mpssas_complete_tm_request(). This allows us to optionally serialize task management commands. Also, change things so that the response to a task management command always comes back through the callback. (Before it could come via the callback or the return value.) mpsvar.h: Add softc variables for the list of active task management commands, the number of active commands, and whether we should allow multiple active task management commands. Add an active command flag. mps.4: Describe the new sysctl/loader tunable variable. Sponsored by: Spectra Logic Corporation
2010-10-07 21:56:10 +00:00
device_get_unit(sc->mps_dev));
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
TUNABLE_INT_FETCH(tmpstr, &sc->disable_msix);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
snprintf(tmpstr, sizeof(tmpstr), "dev.mps.%d.disable_msi",
device_get_unit(sc->mps_dev));
TUNABLE_INT_FETCH(tmpstr, &sc->disable_msi);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
snprintf(tmpstr, sizeof(tmpstr), "dev.mps.%d.max_chains",
device_get_unit(sc->mps_dev));
TUNABLE_INT_FETCH(tmpstr, &sc->max_chains);
Sometimes a device misbehaves so badly that it disrupts the entire system. Add a tunable that allows such a device to be excluded from the driver. The id parameter is the target id that the driver assigns to a given device. dev.mps.X.exclude_ids=<id>,<id> Obtained from: Netflix MFC after: 3 days
2013-08-09 01:09:02 +00:00
bzero(sc->exclude_ids, sizeof(sc->exclude_ids));
snprintf(tmpstr, sizeof(tmpstr), "dev.mps.%d.exclude_ids",
device_get_unit(sc->mps_dev));
TUNABLE_STR_FETCH(tmpstr, sc->exclude_ids, sizeof(sc->exclude_ids));
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
}
static void
mps_setup_sysctl(struct mps_softc *sc)
{
struct sysctl_ctx_list *sysctl_ctx = NULL;
struct sysctl_oid *sysctl_tree = NULL;
char tmpstr[80], tmpstr2[80];
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
/*
* Setup the sysctl variable so the user can change the debug level
* on the fly.
*/
snprintf(tmpstr, sizeof(tmpstr), "MPS controller %d",
device_get_unit(sc->mps_dev));
snprintf(tmpstr2, sizeof(tmpstr2), "%d", device_get_unit(sc->mps_dev));
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
sysctl_ctx = device_get_sysctl_ctx(sc->mps_dev);
if (sysctl_ctx != NULL)
sysctl_tree = device_get_sysctl_tree(sc->mps_dev);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
if (sysctl_tree == NULL) {
sysctl_ctx_init(&sc->sysctl_ctx);
sc->sysctl_tree = SYSCTL_ADD_NODE(&sc->sysctl_ctx,
SYSCTL_STATIC_CHILDREN(_hw_mps), OID_AUTO, tmpstr2,
CTLFLAG_RD, 0, tmpstr);
if (sc->sysctl_tree == NULL)
return;
sysctl_ctx = &sc->sysctl_ctx;
sysctl_tree = sc->sysctl_tree;
}
SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree),
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
OID_AUTO, "debug_level", CTLFLAG_RW, &sc->mps_debug, 0,
"mps debug level");
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree),
OID_AUTO, "disable_msix", CTLFLAG_RD, &sc->disable_msix, 0,
"Disable the use of MSI-X interrupts");
SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree),
OID_AUTO, "disable_msi", CTLFLAG_RD, &sc->disable_msi, 0,
"Disable the use of MSI interrupts");
SYSCTL_ADD_STRING(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree),
OID_AUTO, "firmware_version", CTLFLAG_RW, &sc->fw_version,
strlen(sc->fw_version), "firmware version");
SYSCTL_ADD_STRING(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree),
OID_AUTO, "driver_version", CTLFLAG_RW, MPS_DRIVER_VERSION,
strlen(MPS_DRIVER_VERSION), "driver version");
Turn on serialization of task management commands going down to the controller, but make it optional. After a problem report from Andrew Boyer, it looks like the LSI chip may have issues (the watchdog timer fired) if too many aborts are sent down to the chip at the same time. We know that task management commands are serialized, and although the manual doesn't say it, it may be a good idea to just send one at a time. But, since I'm not certain that this is necessary, add a tunable and sysctl variable (hw.mps.%d.allow_multiple_tm_cmds) to control the driver's behavior. mps.c: Add support for the sysctl and tunable, and add a comment about the possible return values to mps_map_command(). mps_sas.c: Run all task management commands through two new routines, mpssas_issue_tm_request() and mpssas_complete_tm_request(). This allows us to optionally serialize task management commands. Also, change things so that the response to a task management command always comes back through the callback. (Before it could come via the callback or the return value.) mpsvar.h: Add softc variables for the list of active task management commands, the number of active commands, and whether we should allow multiple active task management commands. Add an active command flag. mps.4: Describe the new sysctl/loader tunable variable. Sponsored by: Spectra Logic Corporation
2010-10-07 21:56:10 +00:00
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree),
Fix several issues with the mps(4) driver. When the driver ran out of DMA chaining buffers, it kept the timeout for the I/O, and I/O would stall. The driver was not freezing the device queue on errors. mps.c: Pull command completion logic into a separate function, and call the callback/wakeup for commands that are never sent due to lack of chain buffers. Add a number of extra diagnostic sysctl variables. Handle pre-hardware errors for configuration I/O. This doesn't panic the system, but it will fail the configuration I/O and there is no retry mechanism. So the device probe will not succeed. This should be a very uncommon situation, however. mps_sas.c: Freeze the SIM queue when we run out of chain buffers, and unfreeze it when more commands complete. Freeze the device queue when errors occur, so that CAM can insure proper command ordering. Report pre-hardware errors for task management commands. In general, that shouldn't be possible because task management commands don't have S/G lists, and that is currently the only error path before we get to the hardware. Handle pre-hardware errors (like out of chain elements) for SMP requests. That shouldn't happen either, since we should have enough space for two S/G elements in the standard request. For commands that end with MPI2_IOCSTATUS_SCSI_IOC_TERMINATED and MPI2_IOCSTATUS_SCSI_EXT_TERMINATED, return them with CAM_REQUEUE_REQ to retry them unconditionally. These seem to be related to back end, transport related problems that are hopefully transient. We don't want to go through the retry count for something that is not a permanent error. Keep track of the number of outstanding I/Os. mpsvar.h: Track the number of free chain elements. Add variables for the number of outstanding I/Os, and I/O high water mark. Add variables to track the number of free chain buffers and the chain low water mark, as well as the number of chain allocation failures. Add I/O state flags and an attach done flag. MFC after: 3 days
2011-02-18 17:06:06 +00:00
OID_AUTO, "io_cmds_active", CTLFLAG_RD,
&sc->io_cmds_active, 0, "number of currently active commands");
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree),
Fix several issues with the mps(4) driver. When the driver ran out of DMA chaining buffers, it kept the timeout for the I/O, and I/O would stall. The driver was not freezing the device queue on errors. mps.c: Pull command completion logic into a separate function, and call the callback/wakeup for commands that are never sent due to lack of chain buffers. Add a number of extra diagnostic sysctl variables. Handle pre-hardware errors for configuration I/O. This doesn't panic the system, but it will fail the configuration I/O and there is no retry mechanism. So the device probe will not succeed. This should be a very uncommon situation, however. mps_sas.c: Freeze the SIM queue when we run out of chain buffers, and unfreeze it when more commands complete. Freeze the device queue when errors occur, so that CAM can insure proper command ordering. Report pre-hardware errors for task management commands. In general, that shouldn't be possible because task management commands don't have S/G lists, and that is currently the only error path before we get to the hardware. Handle pre-hardware errors (like out of chain elements) for SMP requests. That shouldn't happen either, since we should have enough space for two S/G elements in the standard request. For commands that end with MPI2_IOCSTATUS_SCSI_IOC_TERMINATED and MPI2_IOCSTATUS_SCSI_EXT_TERMINATED, return them with CAM_REQUEUE_REQ to retry them unconditionally. These seem to be related to back end, transport related problems that are hopefully transient. We don't want to go through the retry count for something that is not a permanent error. Keep track of the number of outstanding I/Os. mpsvar.h: Track the number of free chain elements. Add variables for the number of outstanding I/Os, and I/O high water mark. Add variables to track the number of free chain buffers and the chain low water mark, as well as the number of chain allocation failures. Add I/O state flags and an attach done flag. MFC after: 3 days
2011-02-18 17:06:06 +00:00
OID_AUTO, "io_cmds_highwater", CTLFLAG_RD,
&sc->io_cmds_highwater, 0, "maximum active commands seen");
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree),
Fix several issues with the mps(4) driver. When the driver ran out of DMA chaining buffers, it kept the timeout for the I/O, and I/O would stall. The driver was not freezing the device queue on errors. mps.c: Pull command completion logic into a separate function, and call the callback/wakeup for commands that are never sent due to lack of chain buffers. Add a number of extra diagnostic sysctl variables. Handle pre-hardware errors for configuration I/O. This doesn't panic the system, but it will fail the configuration I/O and there is no retry mechanism. So the device probe will not succeed. This should be a very uncommon situation, however. mps_sas.c: Freeze the SIM queue when we run out of chain buffers, and unfreeze it when more commands complete. Freeze the device queue when errors occur, so that CAM can insure proper command ordering. Report pre-hardware errors for task management commands. In general, that shouldn't be possible because task management commands don't have S/G lists, and that is currently the only error path before we get to the hardware. Handle pre-hardware errors (like out of chain elements) for SMP requests. That shouldn't happen either, since we should have enough space for two S/G elements in the standard request. For commands that end with MPI2_IOCSTATUS_SCSI_IOC_TERMINATED and MPI2_IOCSTATUS_SCSI_EXT_TERMINATED, return them with CAM_REQUEUE_REQ to retry them unconditionally. These seem to be related to back end, transport related problems that are hopefully transient. We don't want to go through the retry count for something that is not a permanent error. Keep track of the number of outstanding I/Os. mpsvar.h: Track the number of free chain elements. Add variables for the number of outstanding I/Os, and I/O high water mark. Add variables to track the number of free chain buffers and the chain low water mark, as well as the number of chain allocation failures. Add I/O state flags and an attach done flag. MFC after: 3 days
2011-02-18 17:06:06 +00:00
OID_AUTO, "chain_free", CTLFLAG_RD,
&sc->chain_free, 0, "number of free chain elements");
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree),
Fix several issues with the mps(4) driver. When the driver ran out of DMA chaining buffers, it kept the timeout for the I/O, and I/O would stall. The driver was not freezing the device queue on errors. mps.c: Pull command completion logic into a separate function, and call the callback/wakeup for commands that are never sent due to lack of chain buffers. Add a number of extra diagnostic sysctl variables. Handle pre-hardware errors for configuration I/O. This doesn't panic the system, but it will fail the configuration I/O and there is no retry mechanism. So the device probe will not succeed. This should be a very uncommon situation, however. mps_sas.c: Freeze the SIM queue when we run out of chain buffers, and unfreeze it when more commands complete. Freeze the device queue when errors occur, so that CAM can insure proper command ordering. Report pre-hardware errors for task management commands. In general, that shouldn't be possible because task management commands don't have S/G lists, and that is currently the only error path before we get to the hardware. Handle pre-hardware errors (like out of chain elements) for SMP requests. That shouldn't happen either, since we should have enough space for two S/G elements in the standard request. For commands that end with MPI2_IOCSTATUS_SCSI_IOC_TERMINATED and MPI2_IOCSTATUS_SCSI_EXT_TERMINATED, return them with CAM_REQUEUE_REQ to retry them unconditionally. These seem to be related to back end, transport related problems that are hopefully transient. We don't want to go through the retry count for something that is not a permanent error. Keep track of the number of outstanding I/Os. mpsvar.h: Track the number of free chain elements. Add variables for the number of outstanding I/Os, and I/O high water mark. Add variables to track the number of free chain buffers and the chain low water mark, as well as the number of chain allocation failures. Add I/O state flags and an attach done flag. MFC after: 3 days
2011-02-18 17:06:06 +00:00
OID_AUTO, "chain_free_lowwater", CTLFLAG_RD,
&sc->chain_free_lowwater, 0,"lowest number of free chain elements");
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
SYSCTL_ADD_INT(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree),
OID_AUTO, "max_chains", CTLFLAG_RD,
&sc->max_chains, 0,"maximum chain frames that will be allocated");
#if __FreeBSD_version >= 900030
SYSCTL_ADD_UQUAD(sysctl_ctx, SYSCTL_CHILDREN(sysctl_tree),
Fix several issues with the mps(4) driver. When the driver ran out of DMA chaining buffers, it kept the timeout for the I/O, and I/O would stall. The driver was not freezing the device queue on errors. mps.c: Pull command completion logic into a separate function, and call the callback/wakeup for commands that are never sent due to lack of chain buffers. Add a number of extra diagnostic sysctl variables. Handle pre-hardware errors for configuration I/O. This doesn't panic the system, but it will fail the configuration I/O and there is no retry mechanism. So the device probe will not succeed. This should be a very uncommon situation, however. mps_sas.c: Freeze the SIM queue when we run out of chain buffers, and unfreeze it when more commands complete. Freeze the device queue when errors occur, so that CAM can insure proper command ordering. Report pre-hardware errors for task management commands. In general, that shouldn't be possible because task management commands don't have S/G lists, and that is currently the only error path before we get to the hardware. Handle pre-hardware errors (like out of chain elements) for SMP requests. That shouldn't happen either, since we should have enough space for two S/G elements in the standard request. For commands that end with MPI2_IOCSTATUS_SCSI_IOC_TERMINATED and MPI2_IOCSTATUS_SCSI_EXT_TERMINATED, return them with CAM_REQUEUE_REQ to retry them unconditionally. These seem to be related to back end, transport related problems that are hopefully transient. We don't want to go through the retry count for something that is not a permanent error. Keep track of the number of outstanding I/Os. mpsvar.h: Track the number of free chain elements. Add variables for the number of outstanding I/Os, and I/O high water mark. Add variables to track the number of free chain buffers and the chain low water mark, as well as the number of chain allocation failures. Add I/O state flags and an attach done flag. MFC after: 3 days
2011-02-18 17:06:06 +00:00
OID_AUTO, "chain_alloc_fail", CTLFLAG_RD,
&sc->chain_alloc_fail, "chain allocation failures");
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
#endif //FreeBSD_version >= 900030
}
Fix several issues with the mps(4) driver. When the driver ran out of DMA chaining buffers, it kept the timeout for the I/O, and I/O would stall. The driver was not freezing the device queue on errors. mps.c: Pull command completion logic into a separate function, and call the callback/wakeup for commands that are never sent due to lack of chain buffers. Add a number of extra diagnostic sysctl variables. Handle pre-hardware errors for configuration I/O. This doesn't panic the system, but it will fail the configuration I/O and there is no retry mechanism. So the device probe will not succeed. This should be a very uncommon situation, however. mps_sas.c: Freeze the SIM queue when we run out of chain buffers, and unfreeze it when more commands complete. Freeze the device queue when errors occur, so that CAM can insure proper command ordering. Report pre-hardware errors for task management commands. In general, that shouldn't be possible because task management commands don't have S/G lists, and that is currently the only error path before we get to the hardware. Handle pre-hardware errors (like out of chain elements) for SMP requests. That shouldn't happen either, since we should have enough space for two S/G elements in the standard request. For commands that end with MPI2_IOCSTATUS_SCSI_IOC_TERMINATED and MPI2_IOCSTATUS_SCSI_EXT_TERMINATED, return them with CAM_REQUEUE_REQ to retry them unconditionally. These seem to be related to back end, transport related problems that are hopefully transient. We don't want to go through the retry count for something that is not a permanent error. Keep track of the number of outstanding I/Os. mpsvar.h: Track the number of free chain elements. Add variables for the number of outstanding I/Os, and I/O high water mark. Add variables to track the number of free chain buffers and the chain low water mark, as well as the number of chain allocation failures. Add I/O state flags and an attach done flag. MFC after: 3 days
2011-02-18 17:06:06 +00:00
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
int
mps_attach(struct mps_softc *sc)
{
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
int error;
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
mps_get_tunables(sc);
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
MPS_FUNCTRACE(sc);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
mtx_init(&sc->mps_mtx, "MPT2SAS lock", NULL, MTX_DEF);
callout_init_mtx(&sc->periodic, &sc->mps_mtx, 0);
TAILQ_INIT(&sc->event_list);
if ((error = mps_transition_ready(sc)) != 0) {
mps_printf(sc, "%s failed to transition ready\n", __func__);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
return (error);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
}
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
sc->facts = malloc(sizeof(MPI2_IOC_FACTS_REPLY), M_MPT2,
M_ZERO|M_NOWAIT);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
if(!sc->facts) {
device_printf(sc->mps_dev, "Cannot allocate memory %s %d\n",
__func__, __LINE__);
return (ENOMEM);
}
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
/*
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
* Get IOC Facts and allocate all structures based on this information.
* A Diag Reset will also call mps_iocfacts_allocate and re-read the IOC
* Facts. If relevant values have changed in IOC Facts, this function
* will free all of the memory based on IOC Facts and reallocate that
* memory. If this fails, any allocated memory should already be freed.
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
*/
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
if ((error = mps_iocfacts_allocate(sc, TRUE)) != 0) {
mps_dprint(sc, MPS_FAULT, "%s IOC Facts based allocation "
"failed with error %d\n", __func__, error);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
return (error);
}
/* Start the periodic watchdog check on the IOC Doorbell */
mps_periodic(sc);
/*
* The portenable will kick off discovery events that will drive the
* rest of the initialization process. The CAM/SAS module will
* hold up the boot sequence until discovery is complete.
*/
sc->mps_ich.ich_func = mps_startup;
sc->mps_ich.ich_arg = sc;
if (config_intrhook_establish(&sc->mps_ich) != 0) {
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_ERROR, "Cannot establish MPS config hook\n");
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
error = EINVAL;
}
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
/*
* Allow IR to shutdown gracefully when shutdown occurs.
*/
sc->shutdown_eh = EVENTHANDLER_REGISTER(shutdown_final,
mpssas_ir_shutdown, sc, SHUTDOWN_PRI_DEFAULT);
if (sc->shutdown_eh == NULL)
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_ERROR, "shutdown event registration "
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
"failed\n");
mps_setup_sysctl(sc);
Fix several issues with the mps(4) driver. When the driver ran out of DMA chaining buffers, it kept the timeout for the I/O, and I/O would stall. The driver was not freezing the device queue on errors. mps.c: Pull command completion logic into a separate function, and call the callback/wakeup for commands that are never sent due to lack of chain buffers. Add a number of extra diagnostic sysctl variables. Handle pre-hardware errors for configuration I/O. This doesn't panic the system, but it will fail the configuration I/O and there is no retry mechanism. So the device probe will not succeed. This should be a very uncommon situation, however. mps_sas.c: Freeze the SIM queue when we run out of chain buffers, and unfreeze it when more commands complete. Freeze the device queue when errors occur, so that CAM can insure proper command ordering. Report pre-hardware errors for task management commands. In general, that shouldn't be possible because task management commands don't have S/G lists, and that is currently the only error path before we get to the hardware. Handle pre-hardware errors (like out of chain elements) for SMP requests. That shouldn't happen either, since we should have enough space for two S/G elements in the standard request. For commands that end with MPI2_IOCSTATUS_SCSI_IOC_TERMINATED and MPI2_IOCSTATUS_SCSI_EXT_TERMINATED, return them with CAM_REQUEUE_REQ to retry them unconditionally. These seem to be related to back end, transport related problems that are hopefully transient. We don't want to go through the retry count for something that is not a permanent error. Keep track of the number of outstanding I/Os. mpsvar.h: Track the number of free chain elements. Add variables for the number of outstanding I/Os, and I/O high water mark. Add variables to track the number of free chain buffers and the chain low water mark, as well as the number of chain allocation failures. Add I/O state flags and an attach done flag. MFC after: 3 days
2011-02-18 17:06:06 +00:00
sc->mps_flags |= MPS_FLAGS_ATTACH_DONE;
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
return (error);
}
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
/* Run through any late-start handlers. */
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
static void
mps_startup(void *arg)
{
struct mps_softc *sc;
sc = (struct mps_softc *)arg;
mps_lock(sc);
mps_unmask_intr(sc);
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
/* initialize device mapping tables */
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
mps_base_static_config_pages(sc);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
mps_mapping_initialize(sc);
mpssas_startup(sc);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
mps_unlock(sc);
}
/* Periodic watchdog. Is called with the driver lock already held. */
static void
mps_periodic(void *arg)
{
struct mps_softc *sc;
uint32_t db;
sc = (struct mps_softc *)arg;
if (sc->mps_flags & MPS_FLAGS_SHUTDOWN)
return;
db = mps_regread(sc, MPI2_DOORBELL_OFFSET);
if ((db & MPI2_IOC_STATE_MASK) == MPI2_IOC_STATE_FAULT) {
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_FAULT, "IOC Fault 0x%08x, Resetting\n", db);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
mps_reinit(sc);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
}
callout_reset(&sc->periodic, MPS_PERIODIC_DELAY * hz, mps_periodic, sc);
}
static void
mps_log_evt_handler(struct mps_softc *sc, uintptr_t data,
MPI2_EVENT_NOTIFICATION_REPLY *event)
{
MPI2_EVENT_DATA_LOG_ENTRY_ADDED *entry;
mps_print_event(sc, event);
switch (event->Event) {
case MPI2_EVENT_LOG_DATA:
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_EVENT, "MPI2_EVENT_LOG_DATA:\n");
if (sc->mps_debug & MPS_EVENT)
hexdump(event->EventData, event->EventDataLength, NULL, 0);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
break;
case MPI2_EVENT_LOG_ENTRY_ADDED:
entry = (MPI2_EVENT_DATA_LOG_ENTRY_ADDED *)event->EventData;
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_EVENT, "MPI2_EVENT_LOG_ENTRY_ADDED event "
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
"0x%x Sequence %d:\n", entry->LogEntryQualifier,
entry->LogSequence);
break;
default:
break;
}
return;
}
static int
mps_attach_log(struct mps_softc *sc)
{
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
u32 events[MPI2_EVENT_NOTIFY_EVENTMASK_WORDS];
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
bzero(events, 16);
setbit(events, MPI2_EVENT_LOG_DATA);
setbit(events, MPI2_EVENT_LOG_ENTRY_ADDED);
mps_register_events(sc, events, mps_log_evt_handler, NULL,
&sc->mps_log_eh);
return (0);
}
static int
mps_detach_log(struct mps_softc *sc)
{
if (sc->mps_log_eh != NULL)
mps_deregister_events(sc, sc->mps_log_eh);
return (0);
}
/*
* Free all of the driver resources and detach submodules. Should be called
* without the lock held.
*/
int
mps_free(struct mps_softc *sc)
{
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
int error;
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
/* Turn off the watchdog */
mps_lock(sc);
sc->mps_flags |= MPS_FLAGS_SHUTDOWN;
mps_unlock(sc);
/* Lock must not be held for this */
callout_drain(&sc->periodic);
if (((error = mps_detach_log(sc)) != 0) ||
((error = mps_detach_sas(sc)) != 0))
return (error);
Bring in a number of mps(4) driver fixes from LSI: 1. Fixed timeout specification for the msleep in mps_wait_command(). Added 30 second timeout for mps_wait_command() calls in mps_user.c. 2. Make sure we call mps_detach_user() from the kldunload path. 3. Raid Hotplug behavior change. The driver now removes a volume when it goes to a failed state, so we also need to add volume back to the OS when it goes to opitimal/degraded/online from failed/missing. Handle raid volume add and remove from the IR_Volume event. 4. Added some more debugging information. 5. Replace xpt_async(AC_LOST_DEVICE, path, NULL) with mpssas_rescan_target(). This is to work around a panic in CAM that shows up when adding a drive with a rescan and removing another device from the driver thread with an AC_LOST_DEVICE async notification. This problem was encountered in testing with the LSI sas2ircu utility, which was used to create a RAID volume from physical disks. The driver has to create the RAID volume target and remove the physical disk targets, and triggered a panic in the process. The CAM issue needs to be fully diagnosed and fixed, but this works around the issue for now. 6. Fix some memory initialization issues in mps_free_command(). 7. Resolve the "devq freeze forever" issue. This was caused by the internal read capacity command issued in the non-head version of the driver. When the command completed with an error, the driver wasn't unfreezing thd device queue. The version in head uses the CAM infrastructure for getting the read capacity information, and therefore doesn't have the same issue. 8. Bump the version to 13.00.00.00-fbsd. (this is very close to LSI's internal stable driver 13.00.00.00) Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> MFC after: 3 days
2012-02-09 00:16:12 +00:00
mps_detach_user(sc);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
/* Put the IOC back in the READY state. */
mps_lock(sc);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
if ((error = mps_transition_ready(sc)) != 0) {
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
mps_unlock(sc);
return (error);
}
mps_unlock(sc);
if (sc->facts != NULL)
free(sc->facts, M_MPT2);
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
/*
* Free all buffers that are based on IOC Facts. A Diag Reset may need
* to free these buffers too.
*/
mps_iocfacts_free(sc);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
if (sc->sysctl_tree != NULL)
sysctl_ctx_free(&sc->sysctl_ctx);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
/* Deregister the shutdown function */
if (sc->shutdown_eh != NULL)
EVENTHANDLER_DEREGISTER(shutdown_final, sc->shutdown_eh);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
mtx_destroy(&sc->mps_mtx);
return (0);
}
Fix several issues with the mps(4) driver. When the driver ran out of DMA chaining buffers, it kept the timeout for the I/O, and I/O would stall. The driver was not freezing the device queue on errors. mps.c: Pull command completion logic into a separate function, and call the callback/wakeup for commands that are never sent due to lack of chain buffers. Add a number of extra diagnostic sysctl variables. Handle pre-hardware errors for configuration I/O. This doesn't panic the system, but it will fail the configuration I/O and there is no retry mechanism. So the device probe will not succeed. This should be a very uncommon situation, however. mps_sas.c: Freeze the SIM queue when we run out of chain buffers, and unfreeze it when more commands complete. Freeze the device queue when errors occur, so that CAM can insure proper command ordering. Report pre-hardware errors for task management commands. In general, that shouldn't be possible because task management commands don't have S/G lists, and that is currently the only error path before we get to the hardware. Handle pre-hardware errors (like out of chain elements) for SMP requests. That shouldn't happen either, since we should have enough space for two S/G elements in the standard request. For commands that end with MPI2_IOCSTATUS_SCSI_IOC_TERMINATED and MPI2_IOCSTATUS_SCSI_EXT_TERMINATED, return them with CAM_REQUEUE_REQ to retry them unconditionally. These seem to be related to back end, transport related problems that are hopefully transient. We don't want to go through the retry count for something that is not a permanent error. Keep track of the number of outstanding I/Os. mpsvar.h: Track the number of free chain elements. Add variables for the number of outstanding I/Os, and I/O high water mark. Add variables to track the number of free chain buffers and the chain low water mark, as well as the number of chain allocation failures. Add I/O state flags and an attach done flag. MFC after: 3 days
2011-02-18 17:06:06 +00:00
static __inline void
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_complete_command(struct mps_softc *sc, struct mps_command *cm)
Fix several issues with the mps(4) driver. When the driver ran out of DMA chaining buffers, it kept the timeout for the I/O, and I/O would stall. The driver was not freezing the device queue on errors. mps.c: Pull command completion logic into a separate function, and call the callback/wakeup for commands that are never sent due to lack of chain buffers. Add a number of extra diagnostic sysctl variables. Handle pre-hardware errors for configuration I/O. This doesn't panic the system, but it will fail the configuration I/O and there is no retry mechanism. So the device probe will not succeed. This should be a very uncommon situation, however. mps_sas.c: Freeze the SIM queue when we run out of chain buffers, and unfreeze it when more commands complete. Freeze the device queue when errors occur, so that CAM can insure proper command ordering. Report pre-hardware errors for task management commands. In general, that shouldn't be possible because task management commands don't have S/G lists, and that is currently the only error path before we get to the hardware. Handle pre-hardware errors (like out of chain elements) for SMP requests. That shouldn't happen either, since we should have enough space for two S/G elements in the standard request. For commands that end with MPI2_IOCSTATUS_SCSI_IOC_TERMINATED and MPI2_IOCSTATUS_SCSI_EXT_TERMINATED, return them with CAM_REQUEUE_REQ to retry them unconditionally. These seem to be related to back end, transport related problems that are hopefully transient. We don't want to go through the retry count for something that is not a permanent error. Keep track of the number of outstanding I/Os. mpsvar.h: Track the number of free chain elements. Add variables for the number of outstanding I/Os, and I/O high water mark. Add variables to track the number of free chain buffers and the chain low water mark, as well as the number of chain allocation failures. Add I/O state flags and an attach done flag. MFC after: 3 days
2011-02-18 17:06:06 +00:00
{
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
MPS_FUNCTRACE(sc);
if (cm == NULL) {
mps_dprint(sc, MPS_ERROR, "Completing NULL command\n");
return;
}
Fix several issues with the mps(4) driver. When the driver ran out of DMA chaining buffers, it kept the timeout for the I/O, and I/O would stall. The driver was not freezing the device queue on errors. mps.c: Pull command completion logic into a separate function, and call the callback/wakeup for commands that are never sent due to lack of chain buffers. Add a number of extra diagnostic sysctl variables. Handle pre-hardware errors for configuration I/O. This doesn't panic the system, but it will fail the configuration I/O and there is no retry mechanism. So the device probe will not succeed. This should be a very uncommon situation, however. mps_sas.c: Freeze the SIM queue when we run out of chain buffers, and unfreeze it when more commands complete. Freeze the device queue when errors occur, so that CAM can insure proper command ordering. Report pre-hardware errors for task management commands. In general, that shouldn't be possible because task management commands don't have S/G lists, and that is currently the only error path before we get to the hardware. Handle pre-hardware errors (like out of chain elements) for SMP requests. That shouldn't happen either, since we should have enough space for two S/G elements in the standard request. For commands that end with MPI2_IOCSTATUS_SCSI_IOC_TERMINATED and MPI2_IOCSTATUS_SCSI_EXT_TERMINATED, return them with CAM_REQUEUE_REQ to retry them unconditionally. These seem to be related to back end, transport related problems that are hopefully transient. We don't want to go through the retry count for something that is not a permanent error. Keep track of the number of outstanding I/Os. mpsvar.h: Track the number of free chain elements. Add variables for the number of outstanding I/Os, and I/O high water mark. Add variables to track the number of free chain buffers and the chain low water mark, as well as the number of chain allocation failures. Add I/O state flags and an attach done flag. MFC after: 3 days
2011-02-18 17:06:06 +00:00
if (cm->cm_flags & MPS_CM_FLAGS_POLLED)
cm->cm_flags |= MPS_CM_FLAGS_COMPLETE;
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
if (cm->cm_complete != NULL) {
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_TRACE,
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
"%s cm %p calling cm_complete %p data %p reply %p\n",
__func__, cm, cm->cm_complete, cm->cm_complete_data,
cm->cm_reply);
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
cm->cm_complete(sc, cm);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
}
Fix several issues with the mps(4) driver. When the driver ran out of DMA chaining buffers, it kept the timeout for the I/O, and I/O would stall. The driver was not freezing the device queue on errors. mps.c: Pull command completion logic into a separate function, and call the callback/wakeup for commands that are never sent due to lack of chain buffers. Add a number of extra diagnostic sysctl variables. Handle pre-hardware errors for configuration I/O. This doesn't panic the system, but it will fail the configuration I/O and there is no retry mechanism. So the device probe will not succeed. This should be a very uncommon situation, however. mps_sas.c: Freeze the SIM queue when we run out of chain buffers, and unfreeze it when more commands complete. Freeze the device queue when errors occur, so that CAM can insure proper command ordering. Report pre-hardware errors for task management commands. In general, that shouldn't be possible because task management commands don't have S/G lists, and that is currently the only error path before we get to the hardware. Handle pre-hardware errors (like out of chain elements) for SMP requests. That shouldn't happen either, since we should have enough space for two S/G elements in the standard request. For commands that end with MPI2_IOCSTATUS_SCSI_IOC_TERMINATED and MPI2_IOCSTATUS_SCSI_EXT_TERMINATED, return them with CAM_REQUEUE_REQ to retry them unconditionally. These seem to be related to back end, transport related problems that are hopefully transient. We don't want to go through the retry count for something that is not a permanent error. Keep track of the number of outstanding I/Os. mpsvar.h: Track the number of free chain elements. Add variables for the number of outstanding I/Os, and I/O high water mark. Add variables to track the number of free chain buffers and the chain low water mark, as well as the number of chain allocation failures. Add I/O state flags and an attach done flag. MFC after: 3 days
2011-02-18 17:06:06 +00:00
if (cm->cm_flags & MPS_CM_FLAGS_WAKEUP) {
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_TRACE, "waking up %p\n", cm);
Fix several issues with the mps(4) driver. When the driver ran out of DMA chaining buffers, it kept the timeout for the I/O, and I/O would stall. The driver was not freezing the device queue on errors. mps.c: Pull command completion logic into a separate function, and call the callback/wakeup for commands that are never sent due to lack of chain buffers. Add a number of extra diagnostic sysctl variables. Handle pre-hardware errors for configuration I/O. This doesn't panic the system, but it will fail the configuration I/O and there is no retry mechanism. So the device probe will not succeed. This should be a very uncommon situation, however. mps_sas.c: Freeze the SIM queue when we run out of chain buffers, and unfreeze it when more commands complete. Freeze the device queue when errors occur, so that CAM can insure proper command ordering. Report pre-hardware errors for task management commands. In general, that shouldn't be possible because task management commands don't have S/G lists, and that is currently the only error path before we get to the hardware. Handle pre-hardware errors (like out of chain elements) for SMP requests. That shouldn't happen either, since we should have enough space for two S/G elements in the standard request. For commands that end with MPI2_IOCSTATUS_SCSI_IOC_TERMINATED and MPI2_IOCSTATUS_SCSI_EXT_TERMINATED, return them with CAM_REQUEUE_REQ to retry them unconditionally. These seem to be related to back end, transport related problems that are hopefully transient. We don't want to go through the retry count for something that is not a permanent error. Keep track of the number of outstanding I/Os. mpsvar.h: Track the number of free chain elements. Add variables for the number of outstanding I/Os, and I/O high water mark. Add variables to track the number of free chain buffers and the chain low water mark, as well as the number of chain allocation failures. Add I/O state flags and an attach done flag. MFC after: 3 days
2011-02-18 17:06:06 +00:00
wakeup(cm);
}
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
if (cm->cm_sc->io_cmds_active != 0) {
cm->cm_sc->io_cmds_active--;
} else {
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_ERROR, "Warning: io_cmds_active is "
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
"out of sync - resynching to 0\n");
}
Fix several issues with the mps(4) driver. When the driver ran out of DMA chaining buffers, it kept the timeout for the I/O, and I/O would stall. The driver was not freezing the device queue on errors. mps.c: Pull command completion logic into a separate function, and call the callback/wakeup for commands that are never sent due to lack of chain buffers. Add a number of extra diagnostic sysctl variables. Handle pre-hardware errors for configuration I/O. This doesn't panic the system, but it will fail the configuration I/O and there is no retry mechanism. So the device probe will not succeed. This should be a very uncommon situation, however. mps_sas.c: Freeze the SIM queue when we run out of chain buffers, and unfreeze it when more commands complete. Freeze the device queue when errors occur, so that CAM can insure proper command ordering. Report pre-hardware errors for task management commands. In general, that shouldn't be possible because task management commands don't have S/G lists, and that is currently the only error path before we get to the hardware. Handle pre-hardware errors (like out of chain elements) for SMP requests. That shouldn't happen either, since we should have enough space for two S/G elements in the standard request. For commands that end with MPI2_IOCSTATUS_SCSI_IOC_TERMINATED and MPI2_IOCSTATUS_SCSI_EXT_TERMINATED, return them with CAM_REQUEUE_REQ to retry them unconditionally. These seem to be related to back end, transport related problems that are hopefully transient. We don't want to go through the retry count for something that is not a permanent error. Keep track of the number of outstanding I/Os. mpsvar.h: Track the number of free chain elements. Add variables for the number of outstanding I/Os, and I/O high water mark. Add variables to track the number of free chain buffers and the chain low water mark, as well as the number of chain allocation failures. Add I/O state flags and an attach done flag. MFC after: 3 days
2011-02-18 17:06:06 +00:00
}
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
static void
mps_sas_log_info(struct mps_softc *sc , u32 log_info)
{
union loginfo_type {
u32 loginfo;
struct {
u32 subcode:16;
u32 code:8;
u32 originator:4;
u32 bus_type:4;
} dw;
};
union loginfo_type sas_loginfo;
char *originator_str = NULL;
sas_loginfo.loginfo = log_info;
if (sas_loginfo.dw.bus_type != 3 /*SAS*/)
return;
/* each nexus loss loginfo */
if (log_info == 0x31170000)
return;
/* eat the loginfos associated with task aborts */
if ((log_info == 30050000 || log_info ==
0x31140000 || log_info == 0x31130000))
return;
switch (sas_loginfo.dw.originator) {
case 0:
originator_str = "IOP";
break;
case 1:
originator_str = "PL";
break;
case 2:
originator_str = "IR";
break;
}
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_LOG, "log_info(0x%08x): originator(%s), "
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
"code(0x%02x), sub_code(0x%04x)\n", log_info,
originator_str, sas_loginfo.dw.code,
sas_loginfo.dw.subcode);
}
static void
mps_display_reply_info(struct mps_softc *sc, uint8_t *reply)
{
MPI2DefaultReply_t *mpi_reply;
u16 sc_status;
mpi_reply = (MPI2DefaultReply_t*)reply;
sc_status = le16toh(mpi_reply->IOCStatus);
if (sc_status & MPI2_IOCSTATUS_FLAG_LOG_INFO_AVAILABLE)
mps_sas_log_info(sc, le32toh(mpi_reply->IOCLogInfo));
}
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
void
mps_intr(void *data)
{
struct mps_softc *sc;
uint32_t status;
sc = (struct mps_softc *)data;
mps_dprint(sc, MPS_TRACE, "%s\n", __func__);
/*
* Check interrupt status register to flush the bus. This is
* needed for both INTx interrupts and driver-driven polling
*/
status = mps_regread(sc, MPI2_HOST_INTERRUPT_STATUS_OFFSET);
if ((status & MPI2_HIS_REPLY_DESCRIPTOR_INTERRUPT) == 0)
return;
mps_lock(sc);
mps_intr_locked(data);
mps_unlock(sc);
return;
}
/*
* In theory, MSI/MSIX interrupts shouldn't need to read any registers on the
* chip. Hopefully this theory is correct.
*/
void
mps_intr_msi(void *data)
{
struct mps_softc *sc;
sc = (struct mps_softc *)data;
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
mps_dprint(sc, MPS_TRACE, "%s\n", __func__);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
mps_lock(sc);
mps_intr_locked(data);
mps_unlock(sc);
return;
}
/*
* The locking is overly broad and simplistic, but easy to deal with for now.
*/
void
mps_intr_locked(void *data)
{
MPI2_REPLY_DESCRIPTORS_UNION *desc;
struct mps_softc *sc;
struct mps_command *cm = NULL;
uint8_t flags;
u_int pq;
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
MPI2_DIAG_RELEASE_REPLY *rel_rep;
mps_fw_diagnostic_buffer_t *pBuffer;
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
sc = (struct mps_softc *)data;
pq = sc->replypostindex;
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
mps_dprint(sc, MPS_TRACE,
"%s sc %p starting with replypostindex %u\n",
__func__, sc, sc->replypostindex);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
for ( ;; ) {
cm = NULL;
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
desc = &sc->post_queue[sc->replypostindex];
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
flags = desc->Default.ReplyFlags &
MPI2_RPY_DESCRIPT_FLAGS_TYPE_MASK;
Fix an event handling bug with the mps(4) driver. This bug manifested itself after repeated device arrivals and departures. The root of the problem was that the last entry in the reply array wasn't initialized/allocated. So every time we got around to that event, we had a bogus address. There were a couple more problems with the code that are also fixed: - The reply mechanism was being treated as sequential (indexed by sc->replycurindex) even though the spec says that the driver should use the ReplyFrameAddress field of the post queue descriptor to figure out where the reply is. There is no guarantee that the reply descriptors will be used in sequential order. - The second word of the reply post queue descriptor wasn't being checked in mps_intr_locked() to make sure that it wasn't 0xffffffff. So the driver could potentially come across a partially DMAed descriptor. - The number of replies allocated was one less than the actual size of the queue. Instead, it was the size of the number of replies that can be used at one time. (Which is one less than the size of the queue.) mps.c: When initializing the entries in the reply free queue, make sure we initialize the full number that we tell the chip we have (sc->fqdepth), not the number that can be used at any one time (sc->num_replies). When allocating replies, make sure we allocate the number of replies that we've told the chip exist, not just the number that can be used simultaneously. Use the ReplyFrameAddress field of the post queue descriptor to figure out which reply is being referenced. This is what the spec says to do, and the spec doesn't guarantee that the replies will be used in order. Put a check in to verify that the reply address passed back from the card is valid. (Panic if it isn't, we'll panic when we try to deference the reply pointer in any case.) In mps_intr_locked(), verify that the second word of the post queue descriptor is not 0xffffffff in addition to verifying that the unused flag is not set, so we can make sure we didn't get a partially DMAed descriptor. Remove references to sc->replycurindex, it isn't needed now. mpsvar.h: Remove replycurindex from the softc, it isn't needed now. Reviewed by: scottl
2010-12-10 21:45:10 +00:00
if ((flags == MPI2_RPY_DESCRIPT_FLAGS_UNUSED)
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
|| (le32toh(desc->Words.High) == 0xffffffff))
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
break;
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
/* increment the replypostindex now, so that event handlers
* and cm completion handlers which decide to do a diag
* reset can zero it without it getting incremented again
* afterwards, and we break out of this loop on the next
* iteration since the reply post queue has been cleared to
* 0xFF and all descriptors look unused (which they are).
*/
if (++sc->replypostindex >= sc->pqdepth)
sc->replypostindex = 0;
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
switch (flags) {
case MPI2_RPY_DESCRIPT_FLAGS_SCSI_IO_SUCCESS:
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
cm = &sc->commands[le16toh(desc->SCSIIOSuccess.SMID)];
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
cm->cm_reply = NULL;
break;
case MPI2_RPY_DESCRIPT_FLAGS_ADDRESS_REPLY:
{
uint32_t baddr;
uint8_t *reply;
Fix an event handling bug with the mps(4) driver. This bug manifested itself after repeated device arrivals and departures. The root of the problem was that the last entry in the reply array wasn't initialized/allocated. So every time we got around to that event, we had a bogus address. There were a couple more problems with the code that are also fixed: - The reply mechanism was being treated as sequential (indexed by sc->replycurindex) even though the spec says that the driver should use the ReplyFrameAddress field of the post queue descriptor to figure out where the reply is. There is no guarantee that the reply descriptors will be used in sequential order. - The second word of the reply post queue descriptor wasn't being checked in mps_intr_locked() to make sure that it wasn't 0xffffffff. So the driver could potentially come across a partially DMAed descriptor. - The number of replies allocated was one less than the actual size of the queue. Instead, it was the size of the number of replies that can be used at one time. (Which is one less than the size of the queue.) mps.c: When initializing the entries in the reply free queue, make sure we initialize the full number that we tell the chip we have (sc->fqdepth), not the number that can be used at any one time (sc->num_replies). When allocating replies, make sure we allocate the number of replies that we've told the chip exist, not just the number that can be used simultaneously. Use the ReplyFrameAddress field of the post queue descriptor to figure out which reply is being referenced. This is what the spec says to do, and the spec doesn't guarantee that the replies will be used in order. Put a check in to verify that the reply address passed back from the card is valid. (Panic if it isn't, we'll panic when we try to deference the reply pointer in any case.) In mps_intr_locked(), verify that the second word of the post queue descriptor is not 0xffffffff in addition to verifying that the unused flag is not set, so we can make sure we didn't get a partially DMAed descriptor. Remove references to sc->replycurindex, it isn't needed now. mpsvar.h: Remove replycurindex from the softc, it isn't needed now. Reviewed by: scottl
2010-12-10 21:45:10 +00:00
/*
* Re-compose the reply address from the address
* sent back from the chip. The ReplyFrameAddress
* is the lower 32 bits of the physical address of
* particular reply frame. Convert that address to
* host format, and then use that to provide the
* offset against the virtual address base
* (sc->reply_frames).
*/
baddr = le32toh(desc->AddressReply.ReplyFrameAddress);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
reply = sc->reply_frames +
Fix an event handling bug with the mps(4) driver. This bug manifested itself after repeated device arrivals and departures. The root of the problem was that the last entry in the reply array wasn't initialized/allocated. So every time we got around to that event, we had a bogus address. There were a couple more problems with the code that are also fixed: - The reply mechanism was being treated as sequential (indexed by sc->replycurindex) even though the spec says that the driver should use the ReplyFrameAddress field of the post queue descriptor to figure out where the reply is. There is no guarantee that the reply descriptors will be used in sequential order. - The second word of the reply post queue descriptor wasn't being checked in mps_intr_locked() to make sure that it wasn't 0xffffffff. So the driver could potentially come across a partially DMAed descriptor. - The number of replies allocated was one less than the actual size of the queue. Instead, it was the size of the number of replies that can be used at one time. (Which is one less than the size of the queue.) mps.c: When initializing the entries in the reply free queue, make sure we initialize the full number that we tell the chip we have (sc->fqdepth), not the number that can be used at any one time (sc->num_replies). When allocating replies, make sure we allocate the number of replies that we've told the chip exist, not just the number that can be used simultaneously. Use the ReplyFrameAddress field of the post queue descriptor to figure out which reply is being referenced. This is what the spec says to do, and the spec doesn't guarantee that the replies will be used in order. Put a check in to verify that the reply address passed back from the card is valid. (Panic if it isn't, we'll panic when we try to deference the reply pointer in any case.) In mps_intr_locked(), verify that the second word of the post queue descriptor is not 0xffffffff in addition to verifying that the unused flag is not set, so we can make sure we didn't get a partially DMAed descriptor. Remove references to sc->replycurindex, it isn't needed now. mpsvar.h: Remove replycurindex from the softc, it isn't needed now. Reviewed by: scottl
2010-12-10 21:45:10 +00:00
(baddr - ((uint32_t)sc->reply_busaddr));
/*
* Make sure the reply we got back is in a valid
* range. If not, go ahead and panic here, since
* we'll probably panic as soon as we deference the
* reply pointer anyway.
*/
if ((reply < sc->reply_frames)
|| (reply > (sc->reply_frames +
(sc->fqdepth * sc->facts->ReplyFrameSize * 4)))) {
printf("%s: WARNING: reply %p out of range!\n",
__func__, reply);
printf("%s: reply_frames %p, fqdepth %d, "
"frame size %d\n", __func__,
sc->reply_frames, sc->fqdepth,
sc->facts->ReplyFrameSize * 4);
printf("%s: baddr %#x,\n", __func__, baddr);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
/* LSI-TODO. See Linux Code. Need Gracefull exit*/
Fix an event handling bug with the mps(4) driver. This bug manifested itself after repeated device arrivals and departures. The root of the problem was that the last entry in the reply array wasn't initialized/allocated. So every time we got around to that event, we had a bogus address. There were a couple more problems with the code that are also fixed: - The reply mechanism was being treated as sequential (indexed by sc->replycurindex) even though the spec says that the driver should use the ReplyFrameAddress field of the post queue descriptor to figure out where the reply is. There is no guarantee that the reply descriptors will be used in sequential order. - The second word of the reply post queue descriptor wasn't being checked in mps_intr_locked() to make sure that it wasn't 0xffffffff. So the driver could potentially come across a partially DMAed descriptor. - The number of replies allocated was one less than the actual size of the queue. Instead, it was the size of the number of replies that can be used at one time. (Which is one less than the size of the queue.) mps.c: When initializing the entries in the reply free queue, make sure we initialize the full number that we tell the chip we have (sc->fqdepth), not the number that can be used at any one time (sc->num_replies). When allocating replies, make sure we allocate the number of replies that we've told the chip exist, not just the number that can be used simultaneously. Use the ReplyFrameAddress field of the post queue descriptor to figure out which reply is being referenced. This is what the spec says to do, and the spec doesn't guarantee that the replies will be used in order. Put a check in to verify that the reply address passed back from the card is valid. (Panic if it isn't, we'll panic when we try to deference the reply pointer in any case.) In mps_intr_locked(), verify that the second word of the post queue descriptor is not 0xffffffff in addition to verifying that the unused flag is not set, so we can make sure we didn't get a partially DMAed descriptor. Remove references to sc->replycurindex, it isn't needed now. mpsvar.h: Remove replycurindex from the softc, it isn't needed now. Reviewed by: scottl
2010-12-10 21:45:10 +00:00
panic("Reply address out of range");
}
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
if (le16toh(desc->AddressReply.SMID) == 0) {
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
if (((MPI2_DEFAULT_REPLY *)reply)->Function ==
MPI2_FUNCTION_DIAG_BUFFER_POST) {
/*
* If SMID is 0 for Diag Buffer Post,
* this implies that the reply is due to
* a release function with a status that
* the buffer has been released. Set
* the buffer flags accordingly.
*/
rel_rep =
(MPI2_DIAG_RELEASE_REPLY *)reply;
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
if (le16toh(rel_rep->IOCStatus) ==
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
MPI2_IOCSTATUS_DIAGNOSTIC_RELEASED)
{
pBuffer =
&sc->fw_diag_buffer_list[
rel_rep->BufferType];
pBuffer->valid_data = TRUE;
pBuffer->owned_by_firmware =
FALSE;
pBuffer->immediate = FALSE;
}
} else
mps_dispatch_event(sc, baddr,
(MPI2_EVENT_NOTIFICATION_REPLY *)
reply);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
} else {
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
cm = &sc->commands[le16toh(desc->AddressReply.SMID)];
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
cm->cm_reply = reply;
cm->cm_reply_data =
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
le32toh(desc->AddressReply.ReplyFrameAddress);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
}
break;
}
case MPI2_RPY_DESCRIPT_FLAGS_TARGETASSIST_SUCCESS:
case MPI2_RPY_DESCRIPT_FLAGS_TARGET_COMMAND_BUFFER:
case MPI2_RPY_DESCRIPT_FLAGS_RAID_ACCELERATOR_SUCCESS:
default:
/* Unhandled */
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_ERROR, "Unhandled reply 0x%x\n",
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
desc->Default.ReplyFlags);
cm = NULL;
break;
}
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
if (cm != NULL) {
// Print Error reply frame
if (cm->cm_reply)
mps_display_reply_info(sc,cm->cm_reply);
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_complete_command(sc, cm);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
}
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
desc->Words.Low = 0xffffffff;
desc->Words.High = 0xffffffff;
}
if (pq != sc->replypostindex) {
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
mps_dprint(sc, MPS_TRACE,
"%s sc %p writing postindex %d\n",
__func__, sc, sc->replypostindex);
mps_regwrite(sc, MPI2_REPLY_POST_HOST_INDEX_OFFSET, sc->replypostindex);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
}
return;
}
static void
mps_dispatch_event(struct mps_softc *sc, uintptr_t data,
MPI2_EVENT_NOTIFICATION_REPLY *reply)
{
struct mps_event_handle *eh;
Fix double ;;
2010-12-06 10:24:06 +00:00
int event, handled = 0;
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
event = le16toh(reply->Event);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
TAILQ_FOREACH(eh, &sc->event_list, eh_list) {
if (isset(eh->mask, event)) {
eh->callback(sc, data, reply);
handled++;
}
}
if (handled == 0)
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_EVENT, "Unhandled event 0x%x\n", le16toh(event));
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
/*
* This is the only place that the event/reply should be freed.
* Anything wanting to hold onto the event data should have
* already copied it into their own storage.
*/
mps_free_reply(sc, data);
}
static void
mps_reregister_events_complete(struct mps_softc *sc, struct mps_command *cm)
{
mps_dprint(sc, MPS_TRACE, "%s\n", __func__);
if (cm->cm_reply)
mps_print_event(sc,
(MPI2_EVENT_NOTIFICATION_REPLY *)cm->cm_reply);
mps_free_command(sc, cm);
/* next, send a port enable */
mpssas_startup(sc);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
}
/*
* For both register_events and update_events, the caller supplies a bitmap
* of events that it _wants_. These functions then turn that into a bitmask
* suitable for the controller.
*/
int
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
mps_register_events(struct mps_softc *sc, u32 *mask,
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
mps_evt_callback_t *cb, void *data, struct mps_event_handle **handle)
{
struct mps_event_handle *eh;
int error = 0;
eh = malloc(sizeof(struct mps_event_handle), M_MPT2, M_WAITOK|M_ZERO);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
if(!eh) {
device_printf(sc->mps_dev, "Cannot allocate memory %s %d\n",
__func__, __LINE__);
return (ENOMEM);
}
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
eh->callback = cb;
eh->data = data;
TAILQ_INSERT_TAIL(&sc->event_list, eh, eh_list);
if (mask != NULL)
error = mps_update_events(sc, eh, mask);
*handle = eh;
return (error);
}
int
mps_update_events(struct mps_softc *sc, struct mps_event_handle *handle,
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
u32 *mask)
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
{
MPI2_EVENT_NOTIFICATION_REQUEST *evtreq;
MPI2_EVENT_NOTIFICATION_REPLY *reply;
struct mps_command *cm;
int error, i;
mps_dprint(sc, MPS_TRACE, "%s\n", __func__);
if ((mask != NULL) && (handle != NULL))
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
bcopy(mask, &handle->mask[0], sizeof(u32) *
MPI2_EVENT_NOTIFY_EVENTMASK_WORDS);
for (i = 0; i < MPI2_EVENT_NOTIFY_EVENTMASK_WORDS; i++)
sc->event_mask[i] = -1;
for (i = 0; i < MPI2_EVENT_NOTIFY_EVENTMASK_WORDS; i++)
sc->event_mask[i] &= ~handle->mask[i];
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
if ((cm = mps_alloc_command(sc)) == NULL)
return (EBUSY);
evtreq = (MPI2_EVENT_NOTIFICATION_REQUEST *)cm->cm_req;
evtreq->Function = MPI2_FUNCTION_EVENT_NOTIFICATION;
evtreq->MsgFlags = 0;
evtreq->SASBroadcastPrimitiveMasks = 0;
#ifdef MPS_DEBUG_ALL_EVENTS
{
u_char fullmask[16];
memset(fullmask, 0x00, 16);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
bcopy(fullmask, &evtreq->EventMasks[0], sizeof(u32) *
MPI2_EVENT_NOTIFY_EVENTMASK_WORDS);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
}
#else
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
for (i = 0; i < MPI2_EVENT_NOTIFY_EVENTMASK_WORDS; i++)
evtreq->EventMasks[i] =
htole32(sc->event_mask[i]);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
#endif
cm->cm_desc.Default.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_DEFAULT_TYPE;
cm->cm_data = NULL;
error = mps_request_polled(sc, cm);
reply = (MPI2_EVENT_NOTIFICATION_REPLY *)cm->cm_reply;
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
if ((reply == NULL) ||
(reply->IOCStatus & MPI2_IOCSTATUS_MASK) != MPI2_IOCSTATUS_SUCCESS)
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
error = ENXIO;
mps_print_event(sc, reply);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
mps_dprint(sc, MPS_TRACE, "%s finished error %d\n", __func__, error);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
mps_free_command(sc, cm);
return (error);
}
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
static int
mps_reregister_events(struct mps_softc *sc)
{
MPI2_EVENT_NOTIFICATION_REQUEST *evtreq;
struct mps_command *cm;
struct mps_event_handle *eh;
int error, i;
mps_dprint(sc, MPS_TRACE, "%s\n", __func__);
/* first, reregister events */
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
for (i = 0; i < MPI2_EVENT_NOTIFY_EVENTMASK_WORDS; i++)
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
sc->event_mask[i] = -1;
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
TAILQ_FOREACH(eh, &sc->event_list, eh_list) {
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
for (i = 0; i < MPI2_EVENT_NOTIFY_EVENTMASK_WORDS; i++)
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
sc->event_mask[i] &= ~eh->mask[i];
}
if ((cm = mps_alloc_command(sc)) == NULL)
return (EBUSY);
evtreq = (MPI2_EVENT_NOTIFICATION_REQUEST *)cm->cm_req;
evtreq->Function = MPI2_FUNCTION_EVENT_NOTIFICATION;
evtreq->MsgFlags = 0;
evtreq->SASBroadcastPrimitiveMasks = 0;
#ifdef MPS_DEBUG_ALL_EVENTS
{
u_char fullmask[16];
memset(fullmask, 0x00, 16);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
bcopy(fullmask, &evtreq->EventMasks[0], sizeof(u32) *
MPI2_EVENT_NOTIFY_EVENTMASK_WORDS);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
}
#else
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
for (i = 0; i < MPI2_EVENT_NOTIFY_EVENTMASK_WORDS; i++)
evtreq->EventMasks[i] =
htole32(sc->event_mask[i]);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
#endif
cm->cm_desc.Default.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_DEFAULT_TYPE;
cm->cm_data = NULL;
cm->cm_complete = mps_reregister_events_complete;
error = mps_map_command(sc, cm);
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
mps_dprint(sc, MPS_TRACE, "%s finished with error %d\n", __func__,
error);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
return (error);
}
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
void
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
mps_deregister_events(struct mps_softc *sc, struct mps_event_handle *handle)
{
TAILQ_REMOVE(&sc->event_list, handle, eh_list);
free(handle, M_MPT2);
}
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
/*
* Add a chain element as the next SGE for the specified command.
* Reset cm_sge and cm_sgesize to indicate all the available space.
*/
static int
mps_add_chain(struct mps_command *cm)
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
{
MPI2_SGE_CHAIN32 *sgc;
struct mps_chain *chain;
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
int space;
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
if (cm->cm_sglsize < MPS_SGC_SIZE)
panic("MPS: Need SGE Error Code\n");
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
chain = mps_alloc_chain(cm->cm_sc);
if (chain == NULL)
return (ENOBUFS);
space = (int)cm->cm_sc->facts->IOCRequestFrameSize * 4;
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
/*
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
* Note: a double-linked list is used to make it easier to
* walk for debugging.
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
*/
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
TAILQ_INSERT_TAIL(&cm->cm_chain_list, chain, chain_link);
sgc = (MPI2_SGE_CHAIN32 *)&cm->cm_sge->MpiChain;
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
sgc->Length = htole16(space);
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
sgc->NextChainOffset = 0;
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
/* TODO Looks like bug in Setting sgc->Flags.
* sgc->Flags = ( MPI2_SGE_FLAGS_CHAIN_ELEMENT | MPI2_SGE_FLAGS_64_BIT_ADDRESSING |
* MPI2_SGE_FLAGS_SYSTEM_ADDRESS) << MPI2_SGE_FLAGS_SHIFT
* This is fine.. because we are not using simple element. In case of
* MPI2_SGE_CHAIN32, we have seperate Length and Flags feild.
*/
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
sgc->Flags = MPI2_SGE_FLAGS_CHAIN_ELEMENT;
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
sgc->Address = htole32(chain->chain_busaddr);
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
cm->cm_sge = (MPI2_SGE_IO_UNION *)&chain->chain->MpiSimple;
cm->cm_sglsize = space;
return (0);
}
/*
* Add one scatter-gather element (chain, simple, transaction context)
* to the scatter-gather list for a command. Maintain cm_sglsize and
* cm_sge as the remaining size and pointer to the next SGE to fill
* in, respectively.
*/
int
mps_push_sge(struct mps_command *cm, void *sgep, size_t len, int segsleft)
{
MPI2_SGE_TRANSACTION_UNION *tc = sgep;
MPI2_SGE_SIMPLE64 *sge = sgep;
int error, type;
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
uint32_t saved_buf_len, saved_address_low, saved_address_high;
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
type = (tc->Flags & MPI2_SGE_FLAGS_ELEMENT_MASK);
#ifdef INVARIANTS
switch (type) {
case MPI2_SGE_FLAGS_TRANSACTION_ELEMENT: {
if (len != tc->DetailsLength + 4)
panic("TC %p length %u or %zu?", tc,
tc->DetailsLength + 4, len);
}
break;
case MPI2_SGE_FLAGS_CHAIN_ELEMENT:
/* Driver only uses 32-bit chain elements */
if (len != MPS_SGC_SIZE)
panic("CHAIN %p length %u or %zu?", sgep,
MPS_SGC_SIZE, len);
break;
case MPI2_SGE_FLAGS_SIMPLE_ELEMENT:
/* Driver only uses 64-bit SGE simple elements */
if (len != MPS_SGE64_SIZE)
panic("SGE simple %p length %u or %zu?", sge,
MPS_SGE64_SIZE, len);
Several fixes to allow firmware/BIOS flash access from user-level: - remove special handling of zero length transfers in mpi_pre_fw_upload(); - add missing MPS_CM_FLAGS_DATAIN flag in mpi_pre_fw_upload(); - move mps_user_setup_request() call into proper place; - increase user command timeout from 30 to 60 seconds; - avoid NULL dereference panic in case of firmware crash. Set max DMA segment size to 24bit, as MPI SGE supports it. Use mps_add_dmaseg() to add empty SGE instead of custom code. Tune endianness safety. Reviewed by: Desai, Kashyap <Kashyap.Desai@lsi.com> Sponsored by: iXsystems, Inc.
2012-08-01 17:31:31 +00:00
if (((le32toh(sge->FlagsLength) >> MPI2_SGE_FLAGS_SHIFT) &
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
MPI2_SGE_FLAGS_ADDRESS_SIZE) == 0)
Several fixes to allow firmware/BIOS flash access from user-level: - remove special handling of zero length transfers in mpi_pre_fw_upload(); - add missing MPS_CM_FLAGS_DATAIN flag in mpi_pre_fw_upload(); - move mps_user_setup_request() call into proper place; - increase user command timeout from 30 to 60 seconds; - avoid NULL dereference panic in case of firmware crash. Set max DMA segment size to 24bit, as MPI SGE supports it. Use mps_add_dmaseg() to add empty SGE instead of custom code. Tune endianness safety. Reviewed by: Desai, Kashyap <Kashyap.Desai@lsi.com> Sponsored by: iXsystems, Inc.
2012-08-01 17:31:31 +00:00
panic("SGE simple %p not marked 64-bit?", sge);
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
break;
default:
panic("Unexpected SGE %p, flags %02x", tc, tc->Flags);
}
#endif
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
/*
* case 1: 1 more segment, enough room for it
* case 2: 2 more segments, enough room for both
* case 3: >=2 more segments, only enough room for 1 and a chain
* case 4: >=1 more segment, enough room for only a chain
* case 5: >=1 more segment, no room for anything (error)
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
*/
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
/*
* There should be room for at least a chain element, or this
* code is buggy. Case (5).
*/
if (cm->cm_sglsize < MPS_SGC_SIZE)
panic("MPS: Need SGE Error Code\n");
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
if (segsleft >= 2 &&
cm->cm_sglsize < len + MPS_SGC_SIZE + MPS_SGE64_SIZE) {
/*
* There are 2 or more segments left to add, and only
* enough room for 1 and a chain. Case (3).
*
* Mark as last element in this chain if necessary.
*/
if (type == MPI2_SGE_FLAGS_SIMPLE_ELEMENT) {
Several fixes to allow firmware/BIOS flash access from user-level: - remove special handling of zero length transfers in mpi_pre_fw_upload(); - add missing MPS_CM_FLAGS_DATAIN flag in mpi_pre_fw_upload(); - move mps_user_setup_request() call into proper place; - increase user command timeout from 30 to 60 seconds; - avoid NULL dereference panic in case of firmware crash. Set max DMA segment size to 24bit, as MPI SGE supports it. Use mps_add_dmaseg() to add empty SGE instead of custom code. Tune endianness safety. Reviewed by: Desai, Kashyap <Kashyap.Desai@lsi.com> Sponsored by: iXsystems, Inc.
2012-08-01 17:31:31 +00:00
sge->FlagsLength |= htole32(
MPI2_SGE_FLAGS_LAST_ELEMENT << MPI2_SGE_FLAGS_SHIFT);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
}
/*
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
* Add the item then a chain. Do the chain now,
* rather than on the next iteration, to simplify
* understanding the code.
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
*/
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
cm->cm_sglsize -= len;
bcopy(sgep, cm->cm_sge, len);
cm->cm_sge = (MPI2_SGE_IO_UNION *)((uintptr_t)cm->cm_sge + len);
return (mps_add_chain(cm));
}
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
if (segsleft >= 1 && cm->cm_sglsize < len + MPS_SGC_SIZE) {
/*
* 1 or more segment, enough room for only a chain.
* Hope the previous element wasn't a Simple entry
* that needed to be marked with
* MPI2_SGE_FLAGS_LAST_ELEMENT. Case (4).
*/
if ((error = mps_add_chain(cm)) != 0)
return (error);
}
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
#ifdef INVARIANTS
/* Case 1: 1 more segment, enough room for it. */
if (segsleft == 1 && cm->cm_sglsize < len)
panic("1 seg left and no room? %u versus %zu",
cm->cm_sglsize, len);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
/* Case 2: 2 more segments, enough room for both */
if (segsleft == 2 && cm->cm_sglsize < len + MPS_SGE64_SIZE)
panic("2 segs left and no room? %u versus %zu",
cm->cm_sglsize, len);
#endif
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
if (segsleft == 1 && type == MPI2_SGE_FLAGS_SIMPLE_ELEMENT) {
/*
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
* If this is a bi-directional request, need to account for that
* here. Save the pre-filled sge values. These will be used
* either for the 2nd SGL or for a single direction SGL. If
* cm_out_len is non-zero, this is a bi-directional request, so
* fill in the OUT SGL first, then the IN SGL, otherwise just
* fill in the IN SGL. Note that at this time, when filling in
* 2 SGL's for a bi-directional request, they both use the same
* DMA buffer (same cm command).
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
*/
Several fixes to allow firmware/BIOS flash access from user-level: - remove special handling of zero length transfers in mpi_pre_fw_upload(); - add missing MPS_CM_FLAGS_DATAIN flag in mpi_pre_fw_upload(); - move mps_user_setup_request() call into proper place; - increase user command timeout from 30 to 60 seconds; - avoid NULL dereference panic in case of firmware crash. Set max DMA segment size to 24bit, as MPI SGE supports it. Use mps_add_dmaseg() to add empty SGE instead of custom code. Tune endianness safety. Reviewed by: Desai, Kashyap <Kashyap.Desai@lsi.com> Sponsored by: iXsystems, Inc.
2012-08-01 17:31:31 +00:00
saved_buf_len = le32toh(sge->FlagsLength) & 0x00FFFFFF;
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
saved_address_low = sge->Address.Low;
saved_address_high = sge->Address.High;
if (cm->cm_out_len) {
Several fixes to allow firmware/BIOS flash access from user-level: - remove special handling of zero length transfers in mpi_pre_fw_upload(); - add missing MPS_CM_FLAGS_DATAIN flag in mpi_pre_fw_upload(); - move mps_user_setup_request() call into proper place; - increase user command timeout from 30 to 60 seconds; - avoid NULL dereference panic in case of firmware crash. Set max DMA segment size to 24bit, as MPI SGE supports it. Use mps_add_dmaseg() to add empty SGE instead of custom code. Tune endianness safety. Reviewed by: Desai, Kashyap <Kashyap.Desai@lsi.com> Sponsored by: iXsystems, Inc.
2012-08-01 17:31:31 +00:00
sge->FlagsLength = htole32(cm->cm_out_len |
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
((uint32_t)(MPI2_SGE_FLAGS_SIMPLE_ELEMENT |
MPI2_SGE_FLAGS_END_OF_BUFFER |
MPI2_SGE_FLAGS_HOST_TO_IOC |
MPI2_SGE_FLAGS_64_BIT_ADDRESSING) <<
Several fixes to allow firmware/BIOS flash access from user-level: - remove special handling of zero length transfers in mpi_pre_fw_upload(); - add missing MPS_CM_FLAGS_DATAIN flag in mpi_pre_fw_upload(); - move mps_user_setup_request() call into proper place; - increase user command timeout from 30 to 60 seconds; - avoid NULL dereference panic in case of firmware crash. Set max DMA segment size to 24bit, as MPI SGE supports it. Use mps_add_dmaseg() to add empty SGE instead of custom code. Tune endianness safety. Reviewed by: Desai, Kashyap <Kashyap.Desai@lsi.com> Sponsored by: iXsystems, Inc.
2012-08-01 17:31:31 +00:00
MPI2_SGE_FLAGS_SHIFT));
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
cm->cm_sglsize -= len;
bcopy(sgep, cm->cm_sge, len);
cm->cm_sge = (MPI2_SGE_IO_UNION *)((uintptr_t)cm->cm_sge
+ len);
}
Several fixes to allow firmware/BIOS flash access from user-level: - remove special handling of zero length transfers in mpi_pre_fw_upload(); - add missing MPS_CM_FLAGS_DATAIN flag in mpi_pre_fw_upload(); - move mps_user_setup_request() call into proper place; - increase user command timeout from 30 to 60 seconds; - avoid NULL dereference panic in case of firmware crash. Set max DMA segment size to 24bit, as MPI SGE supports it. Use mps_add_dmaseg() to add empty SGE instead of custom code. Tune endianness safety. Reviewed by: Desai, Kashyap <Kashyap.Desai@lsi.com> Sponsored by: iXsystems, Inc.
2012-08-01 17:31:31 +00:00
saved_buf_len |=
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
((uint32_t)(MPI2_SGE_FLAGS_SIMPLE_ELEMENT |
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
MPI2_SGE_FLAGS_END_OF_BUFFER |
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
MPI2_SGE_FLAGS_LAST_ELEMENT |
MPI2_SGE_FLAGS_END_OF_LIST |
MPI2_SGE_FLAGS_64_BIT_ADDRESSING) <<
MPI2_SGE_FLAGS_SHIFT);
if (cm->cm_flags & MPS_CM_FLAGS_DATAIN) {
Several fixes to allow firmware/BIOS flash access from user-level: - remove special handling of zero length transfers in mpi_pre_fw_upload(); - add missing MPS_CM_FLAGS_DATAIN flag in mpi_pre_fw_upload(); - move mps_user_setup_request() call into proper place; - increase user command timeout from 30 to 60 seconds; - avoid NULL dereference panic in case of firmware crash. Set max DMA segment size to 24bit, as MPI SGE supports it. Use mps_add_dmaseg() to add empty SGE instead of custom code. Tune endianness safety. Reviewed by: Desai, Kashyap <Kashyap.Desai@lsi.com> Sponsored by: iXsystems, Inc.
2012-08-01 17:31:31 +00:00
saved_buf_len |=
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
((uint32_t)(MPI2_SGE_FLAGS_IOC_TO_HOST) <<
MPI2_SGE_FLAGS_SHIFT);
} else {
Several fixes to allow firmware/BIOS flash access from user-level: - remove special handling of zero length transfers in mpi_pre_fw_upload(); - add missing MPS_CM_FLAGS_DATAIN flag in mpi_pre_fw_upload(); - move mps_user_setup_request() call into proper place; - increase user command timeout from 30 to 60 seconds; - avoid NULL dereference panic in case of firmware crash. Set max DMA segment size to 24bit, as MPI SGE supports it. Use mps_add_dmaseg() to add empty SGE instead of custom code. Tune endianness safety. Reviewed by: Desai, Kashyap <Kashyap.Desai@lsi.com> Sponsored by: iXsystems, Inc.
2012-08-01 17:31:31 +00:00
saved_buf_len |=
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
((uint32_t)(MPI2_SGE_FLAGS_HOST_TO_IOC) <<
MPI2_SGE_FLAGS_SHIFT);
}
Several fixes to allow firmware/BIOS flash access from user-level: - remove special handling of zero length transfers in mpi_pre_fw_upload(); - add missing MPS_CM_FLAGS_DATAIN flag in mpi_pre_fw_upload(); - move mps_user_setup_request() call into proper place; - increase user command timeout from 30 to 60 seconds; - avoid NULL dereference panic in case of firmware crash. Set max DMA segment size to 24bit, as MPI SGE supports it. Use mps_add_dmaseg() to add empty SGE instead of custom code. Tune endianness safety. Reviewed by: Desai, Kashyap <Kashyap.Desai@lsi.com> Sponsored by: iXsystems, Inc.
2012-08-01 17:31:31 +00:00
sge->FlagsLength = htole32(saved_buf_len);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
sge->Address.Low = saved_address_low;
sge->Address.High = saved_address_high;
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
}
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
cm->cm_sglsize -= len;
bcopy(sgep, cm->cm_sge, len);
cm->cm_sge = (MPI2_SGE_IO_UNION *)((uintptr_t)cm->cm_sge + len);
return (0);
}
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
/*
* Add one dma segment to the scatter-gather list for a command.
*/
int
mps_add_dmaseg(struct mps_command *cm, vm_paddr_t pa, size_t len, u_int flags,
int segsleft)
{
MPI2_SGE_SIMPLE64 sge;
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
/*
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
* This driver always uses 64-bit address elements for simplicity.
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
*/
Several fixes to allow firmware/BIOS flash access from user-level: - remove special handling of zero length transfers in mpi_pre_fw_upload(); - add missing MPS_CM_FLAGS_DATAIN flag in mpi_pre_fw_upload(); - move mps_user_setup_request() call into proper place; - increase user command timeout from 30 to 60 seconds; - avoid NULL dereference panic in case of firmware crash. Set max DMA segment size to 24bit, as MPI SGE supports it. Use mps_add_dmaseg() to add empty SGE instead of custom code. Tune endianness safety. Reviewed by: Desai, Kashyap <Kashyap.Desai@lsi.com> Sponsored by: iXsystems, Inc.
2012-08-01 17:31:31 +00:00
bzero(&sge, sizeof(sge));
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
flags |= MPI2_SGE_FLAGS_SIMPLE_ELEMENT |
MPI2_SGE_FLAGS_64_BIT_ADDRESSING;
Several fixes to allow firmware/BIOS flash access from user-level: - remove special handling of zero length transfers in mpi_pre_fw_upload(); - add missing MPS_CM_FLAGS_DATAIN flag in mpi_pre_fw_upload(); - move mps_user_setup_request() call into proper place; - increase user command timeout from 30 to 60 seconds; - avoid NULL dereference panic in case of firmware crash. Set max DMA segment size to 24bit, as MPI SGE supports it. Use mps_add_dmaseg() to add empty SGE instead of custom code. Tune endianness safety. Reviewed by: Desai, Kashyap <Kashyap.Desai@lsi.com> Sponsored by: iXsystems, Inc.
2012-08-01 17:31:31 +00:00
sge.FlagsLength = htole32(len | (flags << MPI2_SGE_FLAGS_SHIFT));
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
mps_from_u64(pa, &sge.Address);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
return (mps_push_sge(cm, &sge, sizeof sge, segsleft));
}
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
static void
mps_data_cb(void *arg, bus_dma_segment_t *segs, int nsegs, int error)
{
struct mps_softc *sc;
struct mps_command *cm;
u_int i, dir, sflags;
cm = (struct mps_command *)arg;
sc = cm->cm_sc;
/*
Add Serial Management Protocol (SMP) passthrough support to CAM. This includes support in the kernel, camcontrol(8), libcam and the mps(4) driver for SMP passthrough. The CAM SCSI probe code has been modified to fetch Inquiry VPD page 0x00 to determine supported pages, and will now fetch page 0x83 in addition to page 0x80 if supported. Add two new CAM CCBs, XPT_SMP_IO, and XPT_GDEV_ADVINFO. The SMP CCB is intended for SMP requests and responses. The ADVINFO is currently used to fetch cached VPD page 0x83 data from the transport layer, but is intended to be extensible to fetch other types of device-specific data. SMP-only devices are not currently represented in the CAM topology, and so the current semantics are that the SIM will route SMP CCBs to either the addressed device, if it contains an SMP target, or its parent, if it contains an SMP target. (This is noted in cam_ccb.h, since it will change later once we have the ability to have SMP-only devices in CAM's topology.) smp_all.c, smp_all.h: New helper routines for SMP. This includes SMP request building routines, response parsing routines, error decoding routines, and structure definitions for a number of SMP commands. libcam/Makefile: Add smp_all.c to libcam, so that SMP functionality is available to userland applications. camcontrol.8, camcontrol.c: Add smp passthrough support to camcontrol. Several new subcommands are now available: 'smpcmd' functions much like 'cmd', except that it allows the user to send generic SMP commands. 'smprg' sends the SMP report general command, and displays the decoded output. It will automatically fetch extended output if it is available. 'smppc' sends the SMP phy control command, with any number of potential options. Among other things, this allows the user to reset a phy on a SAS expander, or disable a phy on an expander. 'smpmaninfo' sends the SMP report manufacturer information and displays the decoded output. 'smpphylist' displays a list of phys on an expander, and the CAM devices attached to those phys, if any. cam.h, cam.c: Add a status value for SMP errors (CAM_SMP_STATUS_ERROR). Add a missing description for CAM_SCSI_IT_NEXUS_LOST. Add support for SMP commands to cam_error_string(). cam_ccb.h: Rename the CAM_DIR_RESV flag to CAM_DIR_BOTH. SMP commands are by nature bi-directional, and we may need to support bi-directional SCSI commands later. Add the XPT_SMP_IO CCB. Since SMP commands are bi-directional, there are pointers for both the request and response. Add a fill routine for SMP CCBs. Add the XPT_GDEV_ADVINFO CCB. This is currently used to fetch cached page 0x83 data from the transport later, but is extensible to fetch many other types of data. cam_periph.c: Add support in cam_periph_mapmem() for XPT_SMP_IO and XPT_GDEV_ADVINFO CCBs. cam_xpt.c: Add support for executing XPT_SMP_IO CCBs. cam_xpt_internal.h: Add fields for VPD pages 0x00 and 0x83 in struct cam_ed. scsi_all.c: Add scsi_get_sas_addr(), a function that parses VPD page 0x83 data and pulls out a SAS address. scsi_all.h: Add VPD page 0x00 and 0x83 structures, and a prototype for scsi_get_sas_addr(). scsi_pass.c: Add support for mapping buffers in XPT_SMP_IO and XPT_GDEV_ADVINFO CCBs. scsi_xpt.c: In the SCSI probe code, first ask the device for VPD page 0x00. If any VPD pages are supported, that page is required to be implemented. Based on the response, we may probe for the serial number (page 0x80) or device id (page 0x83). Add support for the XPT_GDEV_ADVINFO CCB. sys/conf/files: Add smp_all.c. mps.c: Add support for passing in a uio in mps_map_command(), so we can map a S/G list at once. Add support for SMP passthrough commands in mps_data_cb(). SMP is a special case, because the first buffer in the S/G list is outbound and the second buffer is inbound. Add support for warning the user if the busdma code comes back with more buffers than will work for the command. This will, for example, help the user determine why an SMP command failed if busdma comes back with three buffers. mps_pci.c: Add sys/uio.h. mps_sas.c: Add the SAS address and the parent handle to the list of fields we pull from device page 0 and cache in struct mpssas_target. These are needed for SMP passthrough. Add support for the XPT_SMP_IO CCB. For now, this CCB is routed to the addressed device if it supports SMP, or to its parent if it does not and the parent does. This is necessary because CAM does not currently support SMP-only nodes in the topology. Make SMP passthrough support conditional on __FreeBSD_version >= 900026. This will make it easier to MFC this change to the driver without MFCing the CAM changes as well. mps_user.c: Un-staticize mpi_init_sge() so we can use it for the SMP passthrough code. mpsvar.h: Add a uio and iovecs into struct mps_command for SMP passthrough commands. Add a cm_max_segs field to struct mps_command so that we can warn the user if busdma comes back with too many segments. Clear the cm_reply when a command gets freed. If it is not cleared, reply frames will eventually get freed into the pool multiple times and corrupt the pool. (This fix is from scottl.) Add a prototype for mpi_init_sge(). sys/param.h: Bump __FreeBSD_version to 900026 for the for the inclusion of the XPT_GDEV_ADVINFO and XPT_SMP_IO CAM CCBs.
2010-11-30 22:39:46 +00:00
* In this case, just print out a warning and let the chip tell the
* user they did the wrong thing.
*/
if ((cm->cm_max_segs != 0) && (nsegs > cm->cm_max_segs)) {
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_ERROR,
"%s: warning: busdma returned %d segments, "
Add Serial Management Protocol (SMP) passthrough support to CAM. This includes support in the kernel, camcontrol(8), libcam and the mps(4) driver for SMP passthrough. The CAM SCSI probe code has been modified to fetch Inquiry VPD page 0x00 to determine supported pages, and will now fetch page 0x83 in addition to page 0x80 if supported. Add two new CAM CCBs, XPT_SMP_IO, and XPT_GDEV_ADVINFO. The SMP CCB is intended for SMP requests and responses. The ADVINFO is currently used to fetch cached VPD page 0x83 data from the transport layer, but is intended to be extensible to fetch other types of device-specific data. SMP-only devices are not currently represented in the CAM topology, and so the current semantics are that the SIM will route SMP CCBs to either the addressed device, if it contains an SMP target, or its parent, if it contains an SMP target. (This is noted in cam_ccb.h, since it will change later once we have the ability to have SMP-only devices in CAM's topology.) smp_all.c, smp_all.h: New helper routines for SMP. This includes SMP request building routines, response parsing routines, error decoding routines, and structure definitions for a number of SMP commands. libcam/Makefile: Add smp_all.c to libcam, so that SMP functionality is available to userland applications. camcontrol.8, camcontrol.c: Add smp passthrough support to camcontrol. Several new subcommands are now available: 'smpcmd' functions much like 'cmd', except that it allows the user to send generic SMP commands. 'smprg' sends the SMP report general command, and displays the decoded output. It will automatically fetch extended output if it is available. 'smppc' sends the SMP phy control command, with any number of potential options. Among other things, this allows the user to reset a phy on a SAS expander, or disable a phy on an expander. 'smpmaninfo' sends the SMP report manufacturer information and displays the decoded output. 'smpphylist' displays a list of phys on an expander, and the CAM devices attached to those phys, if any. cam.h, cam.c: Add a status value for SMP errors (CAM_SMP_STATUS_ERROR). Add a missing description for CAM_SCSI_IT_NEXUS_LOST. Add support for SMP commands to cam_error_string(). cam_ccb.h: Rename the CAM_DIR_RESV flag to CAM_DIR_BOTH. SMP commands are by nature bi-directional, and we may need to support bi-directional SCSI commands later. Add the XPT_SMP_IO CCB. Since SMP commands are bi-directional, there are pointers for both the request and response. Add a fill routine for SMP CCBs. Add the XPT_GDEV_ADVINFO CCB. This is currently used to fetch cached page 0x83 data from the transport later, but is extensible to fetch many other types of data. cam_periph.c: Add support in cam_periph_mapmem() for XPT_SMP_IO and XPT_GDEV_ADVINFO CCBs. cam_xpt.c: Add support for executing XPT_SMP_IO CCBs. cam_xpt_internal.h: Add fields for VPD pages 0x00 and 0x83 in struct cam_ed. scsi_all.c: Add scsi_get_sas_addr(), a function that parses VPD page 0x83 data and pulls out a SAS address. scsi_all.h: Add VPD page 0x00 and 0x83 structures, and a prototype for scsi_get_sas_addr(). scsi_pass.c: Add support for mapping buffers in XPT_SMP_IO and XPT_GDEV_ADVINFO CCBs. scsi_xpt.c: In the SCSI probe code, first ask the device for VPD page 0x00. If any VPD pages are supported, that page is required to be implemented. Based on the response, we may probe for the serial number (page 0x80) or device id (page 0x83). Add support for the XPT_GDEV_ADVINFO CCB. sys/conf/files: Add smp_all.c. mps.c: Add support for passing in a uio in mps_map_command(), so we can map a S/G list at once. Add support for SMP passthrough commands in mps_data_cb(). SMP is a special case, because the first buffer in the S/G list is outbound and the second buffer is inbound. Add support for warning the user if the busdma code comes back with more buffers than will work for the command. This will, for example, help the user determine why an SMP command failed if busdma comes back with three buffers. mps_pci.c: Add sys/uio.h. mps_sas.c: Add the SAS address and the parent handle to the list of fields we pull from device page 0 and cache in struct mpssas_target. These are needed for SMP passthrough. Add support for the XPT_SMP_IO CCB. For now, this CCB is routed to the addressed device if it supports SMP, or to its parent if it does not and the parent does. This is necessary because CAM does not currently support SMP-only nodes in the topology. Make SMP passthrough support conditional on __FreeBSD_version >= 900026. This will make it easier to MFC this change to the driver without MFCing the CAM changes as well. mps_user.c: Un-staticize mpi_init_sge() so we can use it for the SMP passthrough code. mpsvar.h: Add a uio and iovecs into struct mps_command for SMP passthrough commands. Add a cm_max_segs field to struct mps_command so that we can warn the user if busdma comes back with too many segments. Clear the cm_reply when a command gets freed. If it is not cleared, reply frames will eventually get freed into the pool multiple times and corrupt the pool. (This fix is from scottl.) Add a prototype for mpi_init_sge(). sys/param.h: Bump __FreeBSD_version to 900026 for the for the inclusion of the XPT_GDEV_ADVINFO and XPT_SMP_IO CAM CCBs.
2010-11-30 22:39:46 +00:00
"more than the %d allowed\n", __func__, nsegs,
cm->cm_max_segs);
}
/*
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
* Set up DMA direction flags. Bi-directional requests are also handled
* here. In that case, both direction flags will be set.
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
*/
sflags = 0;
Add Serial Management Protocol (SMP) passthrough support to CAM. This includes support in the kernel, camcontrol(8), libcam and the mps(4) driver for SMP passthrough. The CAM SCSI probe code has been modified to fetch Inquiry VPD page 0x00 to determine supported pages, and will now fetch page 0x83 in addition to page 0x80 if supported. Add two new CAM CCBs, XPT_SMP_IO, and XPT_GDEV_ADVINFO. The SMP CCB is intended for SMP requests and responses. The ADVINFO is currently used to fetch cached VPD page 0x83 data from the transport layer, but is intended to be extensible to fetch other types of device-specific data. SMP-only devices are not currently represented in the CAM topology, and so the current semantics are that the SIM will route SMP CCBs to either the addressed device, if it contains an SMP target, or its parent, if it contains an SMP target. (This is noted in cam_ccb.h, since it will change later once we have the ability to have SMP-only devices in CAM's topology.) smp_all.c, smp_all.h: New helper routines for SMP. This includes SMP request building routines, response parsing routines, error decoding routines, and structure definitions for a number of SMP commands. libcam/Makefile: Add smp_all.c to libcam, so that SMP functionality is available to userland applications. camcontrol.8, camcontrol.c: Add smp passthrough support to camcontrol. Several new subcommands are now available: 'smpcmd' functions much like 'cmd', except that it allows the user to send generic SMP commands. 'smprg' sends the SMP report general command, and displays the decoded output. It will automatically fetch extended output if it is available. 'smppc' sends the SMP phy control command, with any number of potential options. Among other things, this allows the user to reset a phy on a SAS expander, or disable a phy on an expander. 'smpmaninfo' sends the SMP report manufacturer information and displays the decoded output. 'smpphylist' displays a list of phys on an expander, and the CAM devices attached to those phys, if any. cam.h, cam.c: Add a status value for SMP errors (CAM_SMP_STATUS_ERROR). Add a missing description for CAM_SCSI_IT_NEXUS_LOST. Add support for SMP commands to cam_error_string(). cam_ccb.h: Rename the CAM_DIR_RESV flag to CAM_DIR_BOTH. SMP commands are by nature bi-directional, and we may need to support bi-directional SCSI commands later. Add the XPT_SMP_IO CCB. Since SMP commands are bi-directional, there are pointers for both the request and response. Add a fill routine for SMP CCBs. Add the XPT_GDEV_ADVINFO CCB. This is currently used to fetch cached page 0x83 data from the transport later, but is extensible to fetch many other types of data. cam_periph.c: Add support in cam_periph_mapmem() for XPT_SMP_IO and XPT_GDEV_ADVINFO CCBs. cam_xpt.c: Add support for executing XPT_SMP_IO CCBs. cam_xpt_internal.h: Add fields for VPD pages 0x00 and 0x83 in struct cam_ed. scsi_all.c: Add scsi_get_sas_addr(), a function that parses VPD page 0x83 data and pulls out a SAS address. scsi_all.h: Add VPD page 0x00 and 0x83 structures, and a prototype for scsi_get_sas_addr(). scsi_pass.c: Add support for mapping buffers in XPT_SMP_IO and XPT_GDEV_ADVINFO CCBs. scsi_xpt.c: In the SCSI probe code, first ask the device for VPD page 0x00. If any VPD pages are supported, that page is required to be implemented. Based on the response, we may probe for the serial number (page 0x80) or device id (page 0x83). Add support for the XPT_GDEV_ADVINFO CCB. sys/conf/files: Add smp_all.c. mps.c: Add support for passing in a uio in mps_map_command(), so we can map a S/G list at once. Add support for SMP passthrough commands in mps_data_cb(). SMP is a special case, because the first buffer in the S/G list is outbound and the second buffer is inbound. Add support for warning the user if the busdma code comes back with more buffers than will work for the command. This will, for example, help the user determine why an SMP command failed if busdma comes back with three buffers. mps_pci.c: Add sys/uio.h. mps_sas.c: Add the SAS address and the parent handle to the list of fields we pull from device page 0 and cache in struct mpssas_target. These are needed for SMP passthrough. Add support for the XPT_SMP_IO CCB. For now, this CCB is routed to the addressed device if it supports SMP, or to its parent if it does not and the parent does. This is necessary because CAM does not currently support SMP-only nodes in the topology. Make SMP passthrough support conditional on __FreeBSD_version >= 900026. This will make it easier to MFC this change to the driver without MFCing the CAM changes as well. mps_user.c: Un-staticize mpi_init_sge() so we can use it for the SMP passthrough code. mpsvar.h: Add a uio and iovecs into struct mps_command for SMP passthrough commands. Add a cm_max_segs field to struct mps_command so that we can warn the user if busdma comes back with too many segments. Clear the cm_reply when a command gets freed. If it is not cleared, reply frames will eventually get freed into the pool multiple times and corrupt the pool. (This fix is from scottl.) Add a prototype for mpi_init_sge(). sys/param.h: Bump __FreeBSD_version to 900026 for the for the inclusion of the XPT_GDEV_ADVINFO and XPT_SMP_IO CAM CCBs.
2010-11-30 22:39:46 +00:00
if (cm->cm_flags & MPS_CM_FLAGS_SMP_PASS) {
/*
* We have to add a special case for SMP passthrough, there
* is no easy way to generically handle it. The first
* S/G element is used for the command (therefore the
* direction bit needs to be set). The second one is used
* for the reply. We'll leave it to the caller to make
* sure we only have two buffers.
*/
/*
* Even though the busdma man page says it doesn't make
* sense to have both direction flags, it does in this case.
* We have one s/g element being accessed in each direction.
*/
dir = BUS_DMASYNC_PREWRITE | BUS_DMASYNC_PREREAD;
/*
* Set the direction flag on the first buffer in the SMP
* passthrough request. We'll clear it for the second one.
*/
sflags |= MPI2_SGE_FLAGS_DIRECTION |
MPI2_SGE_FLAGS_END_OF_BUFFER;
} else if (cm->cm_flags & MPS_CM_FLAGS_DATAOUT) {
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
sflags |= MPI2_SGE_FLAGS_HOST_TO_IOC;
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
dir = BUS_DMASYNC_PREWRITE;
} else
dir = BUS_DMASYNC_PREREAD;
for (i = 0; i < nsegs; i++) {
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
if ((cm->cm_flags & MPS_CM_FLAGS_SMP_PASS) && (i != 0)) {
Add Serial Management Protocol (SMP) passthrough support to CAM. This includes support in the kernel, camcontrol(8), libcam and the mps(4) driver for SMP passthrough. The CAM SCSI probe code has been modified to fetch Inquiry VPD page 0x00 to determine supported pages, and will now fetch page 0x83 in addition to page 0x80 if supported. Add two new CAM CCBs, XPT_SMP_IO, and XPT_GDEV_ADVINFO. The SMP CCB is intended for SMP requests and responses. The ADVINFO is currently used to fetch cached VPD page 0x83 data from the transport layer, but is intended to be extensible to fetch other types of device-specific data. SMP-only devices are not currently represented in the CAM topology, and so the current semantics are that the SIM will route SMP CCBs to either the addressed device, if it contains an SMP target, or its parent, if it contains an SMP target. (This is noted in cam_ccb.h, since it will change later once we have the ability to have SMP-only devices in CAM's topology.) smp_all.c, smp_all.h: New helper routines for SMP. This includes SMP request building routines, response parsing routines, error decoding routines, and structure definitions for a number of SMP commands. libcam/Makefile: Add smp_all.c to libcam, so that SMP functionality is available to userland applications. camcontrol.8, camcontrol.c: Add smp passthrough support to camcontrol. Several new subcommands are now available: 'smpcmd' functions much like 'cmd', except that it allows the user to send generic SMP commands. 'smprg' sends the SMP report general command, and displays the decoded output. It will automatically fetch extended output if it is available. 'smppc' sends the SMP phy control command, with any number of potential options. Among other things, this allows the user to reset a phy on a SAS expander, or disable a phy on an expander. 'smpmaninfo' sends the SMP report manufacturer information and displays the decoded output. 'smpphylist' displays a list of phys on an expander, and the CAM devices attached to those phys, if any. cam.h, cam.c: Add a status value for SMP errors (CAM_SMP_STATUS_ERROR). Add a missing description for CAM_SCSI_IT_NEXUS_LOST. Add support for SMP commands to cam_error_string(). cam_ccb.h: Rename the CAM_DIR_RESV flag to CAM_DIR_BOTH. SMP commands are by nature bi-directional, and we may need to support bi-directional SCSI commands later. Add the XPT_SMP_IO CCB. Since SMP commands are bi-directional, there are pointers for both the request and response. Add a fill routine for SMP CCBs. Add the XPT_GDEV_ADVINFO CCB. This is currently used to fetch cached page 0x83 data from the transport later, but is extensible to fetch many other types of data. cam_periph.c: Add support in cam_periph_mapmem() for XPT_SMP_IO and XPT_GDEV_ADVINFO CCBs. cam_xpt.c: Add support for executing XPT_SMP_IO CCBs. cam_xpt_internal.h: Add fields for VPD pages 0x00 and 0x83 in struct cam_ed. scsi_all.c: Add scsi_get_sas_addr(), a function that parses VPD page 0x83 data and pulls out a SAS address. scsi_all.h: Add VPD page 0x00 and 0x83 structures, and a prototype for scsi_get_sas_addr(). scsi_pass.c: Add support for mapping buffers in XPT_SMP_IO and XPT_GDEV_ADVINFO CCBs. scsi_xpt.c: In the SCSI probe code, first ask the device for VPD page 0x00. If any VPD pages are supported, that page is required to be implemented. Based on the response, we may probe for the serial number (page 0x80) or device id (page 0x83). Add support for the XPT_GDEV_ADVINFO CCB. sys/conf/files: Add smp_all.c. mps.c: Add support for passing in a uio in mps_map_command(), so we can map a S/G list at once. Add support for SMP passthrough commands in mps_data_cb(). SMP is a special case, because the first buffer in the S/G list is outbound and the second buffer is inbound. Add support for warning the user if the busdma code comes back with more buffers than will work for the command. This will, for example, help the user determine why an SMP command failed if busdma comes back with three buffers. mps_pci.c: Add sys/uio.h. mps_sas.c: Add the SAS address and the parent handle to the list of fields we pull from device page 0 and cache in struct mpssas_target. These are needed for SMP passthrough. Add support for the XPT_SMP_IO CCB. For now, this CCB is routed to the addressed device if it supports SMP, or to its parent if it does not and the parent does. This is necessary because CAM does not currently support SMP-only nodes in the topology. Make SMP passthrough support conditional on __FreeBSD_version >= 900026. This will make it easier to MFC this change to the driver without MFCing the CAM changes as well. mps_user.c: Un-staticize mpi_init_sge() so we can use it for the SMP passthrough code. mpsvar.h: Add a uio and iovecs into struct mps_command for SMP passthrough commands. Add a cm_max_segs field to struct mps_command so that we can warn the user if busdma comes back with too many segments. Clear the cm_reply when a command gets freed. If it is not cleared, reply frames will eventually get freed into the pool multiple times and corrupt the pool. (This fix is from scottl.) Add a prototype for mpi_init_sge(). sys/param.h: Bump __FreeBSD_version to 900026 for the for the inclusion of the XPT_GDEV_ADVINFO and XPT_SMP_IO CAM CCBs.
2010-11-30 22:39:46 +00:00
sflags &= ~MPI2_SGE_FLAGS_DIRECTION;
}
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
error = mps_add_dmaseg(cm, segs[i].ds_addr, segs[i].ds_len,
sflags, nsegs - i);
if (error != 0) {
/* Resource shortage, roll back! */
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_dprint(sc, MPS_INFO, "Out of chain frames, "
"consider increasing hw.mps.max_chains.\n");
Fix several issues with the mps(4) driver. When the driver ran out of DMA chaining buffers, it kept the timeout for the I/O, and I/O would stall. The driver was not freezing the device queue on errors. mps.c: Pull command completion logic into a separate function, and call the callback/wakeup for commands that are never sent due to lack of chain buffers. Add a number of extra diagnostic sysctl variables. Handle pre-hardware errors for configuration I/O. This doesn't panic the system, but it will fail the configuration I/O and there is no retry mechanism. So the device probe will not succeed. This should be a very uncommon situation, however. mps_sas.c: Freeze the SIM queue when we run out of chain buffers, and unfreeze it when more commands complete. Freeze the device queue when errors occur, so that CAM can insure proper command ordering. Report pre-hardware errors for task management commands. In general, that shouldn't be possible because task management commands don't have S/G lists, and that is currently the only error path before we get to the hardware. Handle pre-hardware errors (like out of chain elements) for SMP requests. That shouldn't happen either, since we should have enough space for two S/G elements in the standard request. For commands that end with MPI2_IOCSTATUS_SCSI_IOC_TERMINATED and MPI2_IOCSTATUS_SCSI_EXT_TERMINATED, return them with CAM_REQUEUE_REQ to retry them unconditionally. These seem to be related to back end, transport related problems that are hopefully transient. We don't want to go through the retry count for something that is not a permanent error. Keep track of the number of outstanding I/Os. mpsvar.h: Track the number of free chain elements. Add variables for the number of outstanding I/Os, and I/O high water mark. Add variables to track the number of free chain buffers and the chain low water mark, as well as the number of chain allocation failures. Add I/O state flags and an attach done flag. MFC after: 3 days
2011-02-18 17:06:06 +00:00
cm->cm_flags |= MPS_CM_FLAGS_CHAIN_FAILED;
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
mps_complete_command(sc, cm);
Re-work the internals of adding items to the driver's scatter-gather list. Use the new internals to simplify adding transaction context elements, and in future diffs, more complicated SGLs.
2010-10-14 16:44:05 +00:00
return;
}
}
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
bus_dmamap_sync(sc->buffer_dmat, cm->cm_dmamap, dir);
mps_enqueue_request(sc, cm);
return;
}
Add Serial Management Protocol (SMP) passthrough support to CAM. This includes support in the kernel, camcontrol(8), libcam and the mps(4) driver for SMP passthrough. The CAM SCSI probe code has been modified to fetch Inquiry VPD page 0x00 to determine supported pages, and will now fetch page 0x83 in addition to page 0x80 if supported. Add two new CAM CCBs, XPT_SMP_IO, and XPT_GDEV_ADVINFO. The SMP CCB is intended for SMP requests and responses. The ADVINFO is currently used to fetch cached VPD page 0x83 data from the transport layer, but is intended to be extensible to fetch other types of device-specific data. SMP-only devices are not currently represented in the CAM topology, and so the current semantics are that the SIM will route SMP CCBs to either the addressed device, if it contains an SMP target, or its parent, if it contains an SMP target. (This is noted in cam_ccb.h, since it will change later once we have the ability to have SMP-only devices in CAM's topology.) smp_all.c, smp_all.h: New helper routines for SMP. This includes SMP request building routines, response parsing routines, error decoding routines, and structure definitions for a number of SMP commands. libcam/Makefile: Add smp_all.c to libcam, so that SMP functionality is available to userland applications. camcontrol.8, camcontrol.c: Add smp passthrough support to camcontrol. Several new subcommands are now available: 'smpcmd' functions much like 'cmd', except that it allows the user to send generic SMP commands. 'smprg' sends the SMP report general command, and displays the decoded output. It will automatically fetch extended output if it is available. 'smppc' sends the SMP phy control command, with any number of potential options. Among other things, this allows the user to reset a phy on a SAS expander, or disable a phy on an expander. 'smpmaninfo' sends the SMP report manufacturer information and displays the decoded output. 'smpphylist' displays a list of phys on an expander, and the CAM devices attached to those phys, if any. cam.h, cam.c: Add a status value for SMP errors (CAM_SMP_STATUS_ERROR). Add a missing description for CAM_SCSI_IT_NEXUS_LOST. Add support for SMP commands to cam_error_string(). cam_ccb.h: Rename the CAM_DIR_RESV flag to CAM_DIR_BOTH. SMP commands are by nature bi-directional, and we may need to support bi-directional SCSI commands later. Add the XPT_SMP_IO CCB. Since SMP commands are bi-directional, there are pointers for both the request and response. Add a fill routine for SMP CCBs. Add the XPT_GDEV_ADVINFO CCB. This is currently used to fetch cached page 0x83 data from the transport later, but is extensible to fetch many other types of data. cam_periph.c: Add support in cam_periph_mapmem() for XPT_SMP_IO and XPT_GDEV_ADVINFO CCBs. cam_xpt.c: Add support for executing XPT_SMP_IO CCBs. cam_xpt_internal.h: Add fields for VPD pages 0x00 and 0x83 in struct cam_ed. scsi_all.c: Add scsi_get_sas_addr(), a function that parses VPD page 0x83 data and pulls out a SAS address. scsi_all.h: Add VPD page 0x00 and 0x83 structures, and a prototype for scsi_get_sas_addr(). scsi_pass.c: Add support for mapping buffers in XPT_SMP_IO and XPT_GDEV_ADVINFO CCBs. scsi_xpt.c: In the SCSI probe code, first ask the device for VPD page 0x00. If any VPD pages are supported, that page is required to be implemented. Based on the response, we may probe for the serial number (page 0x80) or device id (page 0x83). Add support for the XPT_GDEV_ADVINFO CCB. sys/conf/files: Add smp_all.c. mps.c: Add support for passing in a uio in mps_map_command(), so we can map a S/G list at once. Add support for SMP passthrough commands in mps_data_cb(). SMP is a special case, because the first buffer in the S/G list is outbound and the second buffer is inbound. Add support for warning the user if the busdma code comes back with more buffers than will work for the command. This will, for example, help the user determine why an SMP command failed if busdma comes back with three buffers. mps_pci.c: Add sys/uio.h. mps_sas.c: Add the SAS address and the parent handle to the list of fields we pull from device page 0 and cache in struct mpssas_target. These are needed for SMP passthrough. Add support for the XPT_SMP_IO CCB. For now, this CCB is routed to the addressed device if it supports SMP, or to its parent if it does not and the parent does. This is necessary because CAM does not currently support SMP-only nodes in the topology. Make SMP passthrough support conditional on __FreeBSD_version >= 900026. This will make it easier to MFC this change to the driver without MFCing the CAM changes as well. mps_user.c: Un-staticize mpi_init_sge() so we can use it for the SMP passthrough code. mpsvar.h: Add a uio and iovecs into struct mps_command for SMP passthrough commands. Add a cm_max_segs field to struct mps_command so that we can warn the user if busdma comes back with too many segments. Clear the cm_reply when a command gets freed. If it is not cleared, reply frames will eventually get freed into the pool multiple times and corrupt the pool. (This fix is from scottl.) Add a prototype for mpi_init_sge(). sys/param.h: Bump __FreeBSD_version to 900026 for the for the inclusion of the XPT_GDEV_ADVINFO and XPT_SMP_IO CAM CCBs.
2010-11-30 22:39:46 +00:00
static void
mps_data_cb2(void *arg, bus_dma_segment_t *segs, int nsegs, bus_size_t mapsize,
int error)
{
mps_data_cb(arg, segs, nsegs, error);
}
Turn on serialization of task management commands going down to the controller, but make it optional. After a problem report from Andrew Boyer, it looks like the LSI chip may have issues (the watchdog timer fired) if too many aborts are sent down to the chip at the same time. We know that task management commands are serialized, and although the manual doesn't say it, it may be a good idea to just send one at a time. But, since I'm not certain that this is necessary, add a tunable and sysctl variable (hw.mps.%d.allow_multiple_tm_cmds) to control the driver's behavior. mps.c: Add support for the sysctl and tunable, and add a comment about the possible return values to mps_map_command(). mps_sas.c: Run all task management commands through two new routines, mpssas_issue_tm_request() and mpssas_complete_tm_request(). This allows us to optionally serialize task management commands. Also, change things so that the response to a task management command always comes back through the callback. (Before it could come via the callback or the return value.) mpsvar.h: Add softc variables for the list of active task management commands, the number of active commands, and whether we should allow multiple active task management commands. Add an active command flag. mps.4: Describe the new sysctl/loader tunable variable. Sponsored by: Spectra Logic Corporation
2010-10-07 21:56:10 +00:00
/*
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
* This is the routine to enqueue commands ansynchronously.
Turn on serialization of task management commands going down to the controller, but make it optional. After a problem report from Andrew Boyer, it looks like the LSI chip may have issues (the watchdog timer fired) if too many aborts are sent down to the chip at the same time. We know that task management commands are serialized, and although the manual doesn't say it, it may be a good idea to just send one at a time. But, since I'm not certain that this is necessary, add a tunable and sysctl variable (hw.mps.%d.allow_multiple_tm_cmds) to control the driver's behavior. mps.c: Add support for the sysctl and tunable, and add a comment about the possible return values to mps_map_command(). mps_sas.c: Run all task management commands through two new routines, mpssas_issue_tm_request() and mpssas_complete_tm_request(). This allows us to optionally serialize task management commands. Also, change things so that the response to a task management command always comes back through the callback. (Before it could come via the callback or the return value.) mpsvar.h: Add softc variables for the list of active task management commands, the number of active commands, and whether we should allow multiple active task management commands. Add an active command flag. mps.4: Describe the new sysctl/loader tunable variable. Sponsored by: Spectra Logic Corporation
2010-10-07 21:56:10 +00:00
* Note that the only error path here is from bus_dmamap_load(), which can
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
* return EINPROGRESS if it is waiting for resources. Other than this, it's
* assumed that if you have a command in-hand, then you have enough credits
* to use it.
Turn on serialization of task management commands going down to the controller, but make it optional. After a problem report from Andrew Boyer, it looks like the LSI chip may have issues (the watchdog timer fired) if too many aborts are sent down to the chip at the same time. We know that task management commands are serialized, and although the manual doesn't say it, it may be a good idea to just send one at a time. But, since I'm not certain that this is necessary, add a tunable and sysctl variable (hw.mps.%d.allow_multiple_tm_cmds) to control the driver's behavior. mps.c: Add support for the sysctl and tunable, and add a comment about the possible return values to mps_map_command(). mps_sas.c: Run all task management commands through two new routines, mpssas_issue_tm_request() and mpssas_complete_tm_request(). This allows us to optionally serialize task management commands. Also, change things so that the response to a task management command always comes back through the callback. (Before it could come via the callback or the return value.) mpsvar.h: Add softc variables for the list of active task management commands, the number of active commands, and whether we should allow multiple active task management commands. Add an active command flag. mps.4: Describe the new sysctl/loader tunable variable. Sponsored by: Spectra Logic Corporation
2010-10-07 21:56:10 +00:00
*/
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
int
mps_map_command(struct mps_softc *sc, struct mps_command *cm)
{
int error = 0;
Add Serial Management Protocol (SMP) passthrough support to CAM. This includes support in the kernel, camcontrol(8), libcam and the mps(4) driver for SMP passthrough. The CAM SCSI probe code has been modified to fetch Inquiry VPD page 0x00 to determine supported pages, and will now fetch page 0x83 in addition to page 0x80 if supported. Add two new CAM CCBs, XPT_SMP_IO, and XPT_GDEV_ADVINFO. The SMP CCB is intended for SMP requests and responses. The ADVINFO is currently used to fetch cached VPD page 0x83 data from the transport layer, but is intended to be extensible to fetch other types of device-specific data. SMP-only devices are not currently represented in the CAM topology, and so the current semantics are that the SIM will route SMP CCBs to either the addressed device, if it contains an SMP target, or its parent, if it contains an SMP target. (This is noted in cam_ccb.h, since it will change later once we have the ability to have SMP-only devices in CAM's topology.) smp_all.c, smp_all.h: New helper routines for SMP. This includes SMP request building routines, response parsing routines, error decoding routines, and structure definitions for a number of SMP commands. libcam/Makefile: Add smp_all.c to libcam, so that SMP functionality is available to userland applications. camcontrol.8, camcontrol.c: Add smp passthrough support to camcontrol. Several new subcommands are now available: 'smpcmd' functions much like 'cmd', except that it allows the user to send generic SMP commands. 'smprg' sends the SMP report general command, and displays the decoded output. It will automatically fetch extended output if it is available. 'smppc' sends the SMP phy control command, with any number of potential options. Among other things, this allows the user to reset a phy on a SAS expander, or disable a phy on an expander. 'smpmaninfo' sends the SMP report manufacturer information and displays the decoded output. 'smpphylist' displays a list of phys on an expander, and the CAM devices attached to those phys, if any. cam.h, cam.c: Add a status value for SMP errors (CAM_SMP_STATUS_ERROR). Add a missing description for CAM_SCSI_IT_NEXUS_LOST. Add support for SMP commands to cam_error_string(). cam_ccb.h: Rename the CAM_DIR_RESV flag to CAM_DIR_BOTH. SMP commands are by nature bi-directional, and we may need to support bi-directional SCSI commands later. Add the XPT_SMP_IO CCB. Since SMP commands are bi-directional, there are pointers for both the request and response. Add a fill routine for SMP CCBs. Add the XPT_GDEV_ADVINFO CCB. This is currently used to fetch cached page 0x83 data from the transport later, but is extensible to fetch many other types of data. cam_periph.c: Add support in cam_periph_mapmem() for XPT_SMP_IO and XPT_GDEV_ADVINFO CCBs. cam_xpt.c: Add support for executing XPT_SMP_IO CCBs. cam_xpt_internal.h: Add fields for VPD pages 0x00 and 0x83 in struct cam_ed. scsi_all.c: Add scsi_get_sas_addr(), a function that parses VPD page 0x83 data and pulls out a SAS address. scsi_all.h: Add VPD page 0x00 and 0x83 structures, and a prototype for scsi_get_sas_addr(). scsi_pass.c: Add support for mapping buffers in XPT_SMP_IO and XPT_GDEV_ADVINFO CCBs. scsi_xpt.c: In the SCSI probe code, first ask the device for VPD page 0x00. If any VPD pages are supported, that page is required to be implemented. Based on the response, we may probe for the serial number (page 0x80) or device id (page 0x83). Add support for the XPT_GDEV_ADVINFO CCB. sys/conf/files: Add smp_all.c. mps.c: Add support for passing in a uio in mps_map_command(), so we can map a S/G list at once. Add support for SMP passthrough commands in mps_data_cb(). SMP is a special case, because the first buffer in the S/G list is outbound and the second buffer is inbound. Add support for warning the user if the busdma code comes back with more buffers than will work for the command. This will, for example, help the user determine why an SMP command failed if busdma comes back with three buffers. mps_pci.c: Add sys/uio.h. mps_sas.c: Add the SAS address and the parent handle to the list of fields we pull from device page 0 and cache in struct mpssas_target. These are needed for SMP passthrough. Add support for the XPT_SMP_IO CCB. For now, this CCB is routed to the addressed device if it supports SMP, or to its parent if it does not and the parent does. This is necessary because CAM does not currently support SMP-only nodes in the topology. Make SMP passthrough support conditional on __FreeBSD_version >= 900026. This will make it easier to MFC this change to the driver without MFCing the CAM changes as well. mps_user.c: Un-staticize mpi_init_sge() so we can use it for the SMP passthrough code. mpsvar.h: Add a uio and iovecs into struct mps_command for SMP passthrough commands. Add a cm_max_segs field to struct mps_command so that we can warn the user if busdma comes back with too many segments. Clear the cm_reply when a command gets freed. If it is not cleared, reply frames will eventually get freed into the pool multiple times and corrupt the pool. (This fix is from scottl.) Add a prototype for mpi_init_sge(). sys/param.h: Bump __FreeBSD_version to 900026 for the for the inclusion of the XPT_GDEV_ADVINFO and XPT_SMP_IO CAM CCBs.
2010-11-30 22:39:46 +00:00
if (cm->cm_flags & MPS_CM_FLAGS_USE_UIO) {
error = bus_dmamap_load_uio(sc->buffer_dmat, cm->cm_dmamap,
&cm->cm_uio, mps_data_cb2, cm, 0);
Reform the busdma API so that new types may be added without modifying every architecture's busdma_machdep.c. It is done by unifying the bus_dmamap_load_buffer() routines so that they may be called from MI code. The MD busdma is then given a chance to do any final processing in the complete() callback. The cam changes unify the bus_dmamap_load* handling in cam drivers. The arm and mips implementations are updated to track virtual addresses for sync(). Previously this was done in a type specific way. Now it is done in a generic way by recording the list of virtuals in the map. Submitted by: jeff (sponsored by EMC/Isilon) Reviewed by: kan (previous version), scottl, mjacob (isp(4), no objections for target mode changes) Discussed with: ian (arm changes) Tested by: marius (sparc64), mips (jmallet), isci(4) on x86 (jharris), amd64 (Fabian Keil <freebsd-listen@fabiankeil.de>)
2013-02-12 16:57:20 +00:00
} else if (cm->cm_flags & MPS_CM_FLAGS_USE_CCB) {
error = bus_dmamap_load_ccb(sc->buffer_dmat, cm->cm_dmamap,
cm->cm_data, mps_data_cb, cm, 0);
Add Serial Management Protocol (SMP) passthrough support to CAM. This includes support in the kernel, camcontrol(8), libcam and the mps(4) driver for SMP passthrough. The CAM SCSI probe code has been modified to fetch Inquiry VPD page 0x00 to determine supported pages, and will now fetch page 0x83 in addition to page 0x80 if supported. Add two new CAM CCBs, XPT_SMP_IO, and XPT_GDEV_ADVINFO. The SMP CCB is intended for SMP requests and responses. The ADVINFO is currently used to fetch cached VPD page 0x83 data from the transport layer, but is intended to be extensible to fetch other types of device-specific data. SMP-only devices are not currently represented in the CAM topology, and so the current semantics are that the SIM will route SMP CCBs to either the addressed device, if it contains an SMP target, or its parent, if it contains an SMP target. (This is noted in cam_ccb.h, since it will change later once we have the ability to have SMP-only devices in CAM's topology.) smp_all.c, smp_all.h: New helper routines for SMP. This includes SMP request building routines, response parsing routines, error decoding routines, and structure definitions for a number of SMP commands. libcam/Makefile: Add smp_all.c to libcam, so that SMP functionality is available to userland applications. camcontrol.8, camcontrol.c: Add smp passthrough support to camcontrol. Several new subcommands are now available: 'smpcmd' functions much like 'cmd', except that it allows the user to send generic SMP commands. 'smprg' sends the SMP report general command, and displays the decoded output. It will automatically fetch extended output if it is available. 'smppc' sends the SMP phy control command, with any number of potential options. Among other things, this allows the user to reset a phy on a SAS expander, or disable a phy on an expander. 'smpmaninfo' sends the SMP report manufacturer information and displays the decoded output. 'smpphylist' displays a list of phys on an expander, and the CAM devices attached to those phys, if any. cam.h, cam.c: Add a status value for SMP errors (CAM_SMP_STATUS_ERROR). Add a missing description for CAM_SCSI_IT_NEXUS_LOST. Add support for SMP commands to cam_error_string(). cam_ccb.h: Rename the CAM_DIR_RESV flag to CAM_DIR_BOTH. SMP commands are by nature bi-directional, and we may need to support bi-directional SCSI commands later. Add the XPT_SMP_IO CCB. Since SMP commands are bi-directional, there are pointers for both the request and response. Add a fill routine for SMP CCBs. Add the XPT_GDEV_ADVINFO CCB. This is currently used to fetch cached page 0x83 data from the transport later, but is extensible to fetch many other types of data. cam_periph.c: Add support in cam_periph_mapmem() for XPT_SMP_IO and XPT_GDEV_ADVINFO CCBs. cam_xpt.c: Add support for executing XPT_SMP_IO CCBs. cam_xpt_internal.h: Add fields for VPD pages 0x00 and 0x83 in struct cam_ed. scsi_all.c: Add scsi_get_sas_addr(), a function that parses VPD page 0x83 data and pulls out a SAS address. scsi_all.h: Add VPD page 0x00 and 0x83 structures, and a prototype for scsi_get_sas_addr(). scsi_pass.c: Add support for mapping buffers in XPT_SMP_IO and XPT_GDEV_ADVINFO CCBs. scsi_xpt.c: In the SCSI probe code, first ask the device for VPD page 0x00. If any VPD pages are supported, that page is required to be implemented. Based on the response, we may probe for the serial number (page 0x80) or device id (page 0x83). Add support for the XPT_GDEV_ADVINFO CCB. sys/conf/files: Add smp_all.c. mps.c: Add support for passing in a uio in mps_map_command(), so we can map a S/G list at once. Add support for SMP passthrough commands in mps_data_cb(). SMP is a special case, because the first buffer in the S/G list is outbound and the second buffer is inbound. Add support for warning the user if the busdma code comes back with more buffers than will work for the command. This will, for example, help the user determine why an SMP command failed if busdma comes back with three buffers. mps_pci.c: Add sys/uio.h. mps_sas.c: Add the SAS address and the parent handle to the list of fields we pull from device page 0 and cache in struct mpssas_target. These are needed for SMP passthrough. Add support for the XPT_SMP_IO CCB. For now, this CCB is routed to the addressed device if it supports SMP, or to its parent if it does not and the parent does. This is necessary because CAM does not currently support SMP-only nodes in the topology. Make SMP passthrough support conditional on __FreeBSD_version >= 900026. This will make it easier to MFC this change to the driver without MFCing the CAM changes as well. mps_user.c: Un-staticize mpi_init_sge() so we can use it for the SMP passthrough code. mpsvar.h: Add a uio and iovecs into struct mps_command for SMP passthrough commands. Add a cm_max_segs field to struct mps_command so that we can warn the user if busdma comes back with too many segments. Clear the cm_reply when a command gets freed. If it is not cleared, reply frames will eventually get freed into the pool multiple times and corrupt the pool. (This fix is from scottl.) Add a prototype for mpi_init_sge(). sys/param.h: Bump __FreeBSD_version to 900026 for the for the inclusion of the XPT_GDEV_ADVINFO and XPT_SMP_IO CAM CCBs.
2010-11-30 22:39:46 +00:00
} else if ((cm->cm_data != NULL) && (cm->cm_length != 0)) {
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
error = bus_dmamap_load(sc->buffer_dmat, cm->cm_dmamap,
cm->cm_data, cm->cm_length, mps_data_cb, cm, 0);
} else {
/* Add a zero-length element as needed */
Several fixes to allow firmware/BIOS flash access from user-level: - remove special handling of zero length transfers in mpi_pre_fw_upload(); - add missing MPS_CM_FLAGS_DATAIN flag in mpi_pre_fw_upload(); - move mps_user_setup_request() call into proper place; - increase user command timeout from 30 to 60 seconds; - avoid NULL dereference panic in case of firmware crash. Set max DMA segment size to 24bit, as MPI SGE supports it. Use mps_add_dmaseg() to add empty SGE instead of custom code. Tune endianness safety. Reviewed by: Desai, Kashyap <Kashyap.Desai@lsi.com> Sponsored by: iXsystems, Inc.
2012-08-01 17:31:31 +00:00
if (cm->cm_sge != NULL)
mps_add_dmaseg(cm, 0, 0, 0, 1);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
mps_enqueue_request(sc, cm);
}
return (error);
}
/*
* This is the routine to enqueue commands synchronously. An error of
* EINPROGRESS from mps_map_command() is ignored since the command will
* be executed and enqueued automatically. Other errors come from msleep().
*/
int
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
mps_wait_command(struct mps_softc *sc, struct mps_command *cm, int timeout,
int sleep_flag)
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
{
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
int error, rc;
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
struct timeval cur_time, start_time;
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
if (sc->mps_flags & MPS_FLAGS_DIAGRESET)
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
return EBUSY;
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
cm->cm_complete = NULL;
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
cm->cm_flags |= (MPS_CM_FLAGS_WAKEUP + MPS_CM_FLAGS_POLLED);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
error = mps_map_command(sc, cm);
if ((error != 0) && (error != EINPROGRESS))
return (error);
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
// Check for context and wait for 50 mSec at a time until time has
// expired or the command has finished. If msleep can't be used, need
// to poll.
if (curthread->td_no_sleeping != 0)
sleep_flag = NO_SLEEP;
getmicrotime(&start_time);
if (mtx_owned(&sc->mps_mtx) && sleep_flag == CAN_SLEEP) {
error = msleep(cm, &sc->mps_mtx, 0, "mpswait", timeout*hz);
} else {
while ((cm->cm_flags & MPS_CM_FLAGS_COMPLETE) == 0) {
mps_intr_locked(sc);
if (sleep_flag == CAN_SLEEP)
pause("mpswait", hz/20);
else
DELAY(50000);
getmicrotime(&cur_time);
if ((cur_time.tv_sec - start_time.tv_sec) > timeout) {
error = EWOULDBLOCK;
break;
}
}
}
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
if (error == EWOULDBLOCK) {
mps_dprint(sc, MPS_FAULT, "Calling Reinit from %s\n", __func__);
rc = mps_reinit(sc);
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
mps_dprint(sc, MPS_FAULT, "Reinit %s\n", (rc == 0) ? "success" :
"failed");
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
error = ETIMEDOUT;
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
}
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
return (error);
}
/*
* This is the routine to enqueue a command synchonously and poll for
* completion. Its use should be rare.
*/
int
mps_request_polled(struct mps_softc *sc, struct mps_command *cm)
{
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
int error, timeout = 0, rc;
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
error = 0;
cm->cm_flags |= MPS_CM_FLAGS_POLLED;
cm->cm_complete = NULL;
mps_map_command(sc, cm);
while ((cm->cm_flags & MPS_CM_FLAGS_COMPLETE) == 0) {
mps_intr_locked(sc);
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
DELAY(50 * 1000);
if (timeout++ > 1000) {
mps_dprint(sc, MPS_FAULT, "polling failed\n");
error = ETIMEDOUT;
break;
}
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
}
Bring in LSI's latest mps(4) 6Gb SAS and WarpDrive driver, version 14.00.00.01-fbsd. Their description of the changes is as follows: 1. Copyright contents has been changed in all respective .c and .h files 2. Support for WRITE12 and READ12 for direct-io (warpdrive only) has been added. 3. Driver has added checks to see if Drive has READ_CAP_16 support before sending it down to the device. If SPC3_SID_PROTECT flag is set in the inquiry data, the device supports protection information, and must support the 16 byte read capacity command, otherwise continue without sending read cap 16. This will optimize driver performance, since it will not send READ_CAP_16 to the drive which does not have support of READ_CAP_16. 4. With new approach, "MPTIOCTL_RESET_ADAPTER" IOCTL will not use DELAY() which is busy loop implementation. It will use <msleep> (Better way to sleep without busy loop). Also from the HBA reset code path and some other places, DELAY() is replaced with msleep() or "pause()", which is based on sleep/wakeup style calls. Driver use msleep()/pause() instead of DELAY based on CAN_SLEEP/NO_SLEEP flags to avoid busy loop which is not required all the time.e.a a. While driver is getting loaded, driver calls most of the commands with NO_SLEEP. b. When Driver is functional and it needs Reinit of HBA, CAN_SLEEP flag is used. 5. <mpslsi> driver is not Endian safe. It will not work on Big Endian machines like Sparc and PowerPC platforms because it assumes it is running on a Little Endian machine. Driver code is modified such way that it does not assume CPU arch is Little Endian. a. All places where Driver interacts from HBA to Host, it converts Little Endian format to CPU format. b. All places where Driver interacts from Host to HBA, it converts CPU format to Little Endian. 6. Findout memory leaks in FreeBSD Driver and resolve those, such as memory leak in targ's luns creation/deletion. Also added additional checks to see memory allocation success/fail. 7. Add loginfo prints as debug message, i.e. When FW sends any loginfo, Driver should print those as debug message. This will help for debugging purpose. 8. There is possibility to get config request timeout. Current driver is able to detect config request timetout, but it does not do anything on config_request timeout. Driver should call mps_reinit() if any request_poll (which is called as part of config_request) is time out. 9. cdb length check is required for 32 byte CDB. Add correct mpi control value for 32 bit CDB as below while submitting SCSI IO Request to controller. mpi_control |= 4 << MPI2_SCSIIO_CONTROL_ADDCDBLEN_SHIFT; 10. Check the actual status of Message unit reset (mps_message_unit_reset).Previously FreeBSD Driver just writes MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET and never check the ack (it just wait for 50 millisecond). So, Driver now check the status of "MPI2_FUNCTION_IOC_MESSAGE_UNIT_RESET" after writing it to the FW. Now it also checking for whether doorbell ack uses msleep with proper sleep flags, instead of <DELAY>. 11. Previously CAM does not detect Multi-Lun Devices. In order to detect Multi-Lun Devices by CAM the driver needs following change set: a. There is "max_lun" field which Driver need to set based on hw/fw support. Currently LSI released driver does not set this field. b. Default of "max_lun" should not be 0 in OS, but it is currently set to 0 in CAM layer. c. Export max_lun capacity to 255 12. Driver will not reset target info after port enable complete and also do Device removal when Device remove from FW. The detail description is as follows a. When Driver receive WD PD add events, it will add all information in driver local data structure. b. Only for WD, we have below checks after port enable completes, where driver clear off all information retrieved at #1. if ((sc->WD_available && (sc->WD_hide_expose == MPS_WD_HIDE_ALWAYS)) || (sc->WD_valid_config && (sc->WD_hide_expose == MPS_WD_HIDE_IF_VOLUME)) { // clear off target data structure. } It is mainly not to attach PDs to OS. FreeBSD does bus rescan as older Parallel scsi style. So Driver needs to handle which Drive is visible to OS. That is a reason we have to clear off targ information for PDs. Again, above logic was implemented long time ago. Similar concept we have for non-wd also. For that, LSI have introduced different logic to hide PDs. Eventually, because of above gap, when Phy goes offline, we observe below failure. That is what Driver is not doing complete removal of device with FW. (which was pointed by Scott) Apr 5 02:39:24 Freebsd7 kernel: mpslsi0: mpssas_prepare_remove Apr 5 02:39:24 Freebsd7 kernel: mpssas_prepare_remove 497 : invalid handle 0xe Now Driver will not reset target info after port enable complete and also will do Device removal when Device remove from FW. 13. Returning "CAM_SEL_TIMEOUT" instead of "CAM_TID_INVALID" error code on request to the Target IDs that have no devices conected at that moment. As if "CAM_TID_INVALID" error code is returned to the CAM Layaer then it results in a huge chain of errors in verbose kernel messages on boot and every hot-plug event. Submitted by: Sreekanth Reddy <Sreekanth.Reddy@lsi.com> MFC after: 3 days
2012-06-28 03:48:54 +00:00
if (error) {
mps_dprint(sc, MPS_FAULT, "Calling Reinit from %s\n", __func__);
rc = mps_reinit(sc);
mps_dprint(sc, MPS_FAULT, "Reinit %s\n",
(rc == 0) ? "success" : "failed");
}
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
return (error);
}
/*
* The MPT driver had a verbose interface for config pages. In this driver,
* reduce it to much simplier terms, similar to the Linux driver.
*/
int
mps_read_config_page(struct mps_softc *sc, struct mps_config_params *params)
{
MPI2_CONFIG_REQUEST *req;
struct mps_command *cm;
int error;
if (sc->mps_flags & MPS_FLAGS_BUSY) {
return (EBUSY);
}
cm = mps_alloc_command(sc);
if (cm == NULL) {
return (EBUSY);
}
req = (MPI2_CONFIG_REQUEST *)cm->cm_req;
req->Function = MPI2_FUNCTION_CONFIG;
req->Action = params->action;
req->SGLFlags = 0;
req->ChainOffset = 0;
req->PageAddress = params->page_address;
sys/dev/mps/mps.c sys/dev/mps/mps_user.c Fix uninitialized memory reference in mps_read_config_page. It was referencing a field (params->hdr.Ext.ExtPageType) that would only be set when reading an Extended config page. The symptom was that MPSIO_READ_CFG_PAGE ioctls would randomly fail with MPI2_IOCSTATUS_CONFIG_INVALID_PAGE errors. The solution is to determine whether an extended or an ordinary config page is requested by looking at the PageType field, which should be available regardless. Similarly, mps_user_read_extcfg_header and mps_user_read_extcfg_page, which call mps_read_config_page, had to be fixed to always set the PageType field. They were implicitly assuming that mps_read_config_page always operated on Extended pages. Reviewed by: ken Approved by: ken (mentor) MFC after: 3 days
2013-06-04 22:32:33 +00:00
if (params->hdr.Struct.PageType == MPI2_CONFIG_PAGETYPE_EXTENDED) {
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
MPI2_CONFIG_EXTENDED_PAGE_HEADER *hdr;
hdr = &params->hdr.Ext;
req->ExtPageType = hdr->ExtPageType;
req->ExtPageLength = hdr->ExtPageLength;
req->Header.PageType = MPI2_CONFIG_PAGETYPE_EXTENDED;
req->Header.PageLength = 0; /* Must be set to zero */
req->Header.PageNumber = hdr->PageNumber;
req->Header.PageVersion = hdr->PageVersion;
} else {
MPI2_CONFIG_PAGE_HEADER *hdr;
hdr = &params->hdr.Struct;
req->Header.PageType = hdr->PageType;
req->Header.PageNumber = hdr->PageNumber;
req->Header.PageLength = hdr->PageLength;
req->Header.PageVersion = hdr->PageVersion;
}
cm->cm_data = params->buffer;
cm->cm_length = params->length;
cm->cm_sge = &req->PageBufferSGE;
cm->cm_sglsize = sizeof(MPI2_SGE_IO_UNION);
cm->cm_flags = MPS_CM_FLAGS_SGE_SIMPLE | MPS_CM_FLAGS_DATAIN;
cm->cm_desc.Default.RequestFlags = MPI2_REQ_DESCRIPT_FLAGS_DEFAULT_TYPE;
Always set cm_complete_data before calling mps_config_complete(). Reviewed by: ken
2010-10-12 19:24:29 +00:00
cm->cm_complete_data = params;
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
if (params->callback != NULL) {
cm->cm_complete = mps_config_complete;
return (mps_map_command(sc, cm));
} else {
Merge in phase 14+ -> 16 mps driver fixes from LSI: --------------------------------------------------------------- System panics during a Port reset with ouststanding I/O --------------------------------------------------------------- It is possible to call mps_mapping_free_memory after this memory is already freed, causing a panic. Removed this extra call to mps_mappiing_free_memory and call mps_mapping_exit in place of the mps_mapping_free_memory call so that any outstanding mapping items can be flushed before memory is freed. --------------------------------------------------------------- Correct memory leak during a Port reset with ouststanding I/O --------------------------------------------------------------- In mps_reinit function, the mapping memory was not being freed before being re-allocated. Added line to call the memory free function for mapping memory. --------------------------------------------------------------- Use CAM_SIM_QUEUED flag in Driver IO path. --------------------------------------------------------------- This flag informs the XPT that successful abort of a CCB requires an abort ccb to be issued to the SIM. While processing SCSI IO's, set the CAM_SIM_QUEUED flag in the status for the IO. When the command completes, clear this flag. --------------------------------------------------------------- Check for CAM_REQ_INPROG in I/O path. --------------------------------------------------------------- Added a check in mpssas_action_scsiio for the In Progress status for the IO. If this flag is set, the IO has already been aborted by the upper layer (before CAM_SIM_QUEUED was set) and there is no need to send the IO. The request will be completed without error. --------------------------------------------------------------- Improve "doorbell handshake method" for mps_get_iocfacts --------------------------------------------------------------- Removed call to get Port Facts since this information is not used currently. Added mps_iocfacts_allocate function to allocate memory that is based on IOC Facts data. Added mps_iocfacts_free function to free memory that is based on IOC Facts data. Both of the functions are used when a Diag Reset is performed or when the driver is attached/detached. This is needed in case IOC Facts changes after a Diag Reset, which could happen if FW is upgraded. Moved call of mps_bases_static_config_pages from the attach routine to after the IOC is ready to process accesses based on the new memory allocations (instead of polling through the Doorbell). --------------------------------------------------------------- Set TimeStamp in INIT message in millisecond format Set the IOC --------------------------------------------------------------- --------------------------------------------------------------- Prefer mps_wait_command to mps_request_polled --------------------------------------------------------------- Instead of using mps_request_polled, call mps_wait_command whenever possible. Change the mps_wait_command function to check the current context and either use interrupt context or poll if required by using the pause or DELAY function. Added a check after waiting 50mSecs to see if the command has timed out. This is only done if polliing, the msleep command will automatically timeout if the command has taken too long to complete. --------------------------------------------------------------- Integrated RAID: Volume Activation Failed error message is displayed though the volume has been activated. --------------------------------------------------------------- Instead of failing an IOCTL request that does not have a large enough buffer to hold the complete reply, copy as much data from the reply as possible into the user's buffer and log a message saying that the user's buffer was smaller than the returned data. --------------------------------------------------------------- mapping_add_new_device failure due to persistent table FULL --------------------------------------------------------------- When a new device is added, if it is determined that the device persistent table is being used and is full, instead of displaying a message for this condition every time, only log a message if the MPS_INFO bit is set in the debug_flags. Submitted by: LSI MFC after: 1 week
2013-07-22 18:41:53 +00:00
error = mps_wait_command(sc, cm, 0, CAN_SLEEP);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
if (error) {
mps_dprint(sc, MPS_FAULT,
"Error %d reading config page\n", error);
mps_free_command(sc, cm);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
return (error);
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
}
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
mps_config_complete(sc, cm);
}
return (0);
}
int
mps_write_config_page(struct mps_softc *sc, struct mps_config_params *params)
{
return (EINVAL);
}
static void
mps_config_complete(struct mps_softc *sc, struct mps_command *cm)
{
MPI2_CONFIG_REPLY *reply;
struct mps_config_params *params;
Overhaul error, information, and debug logging. Obtained from: Netflix MFC after: 3 days
2013-07-19 00:12:41 +00:00
MPS_FUNCTRACE(sc);
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
params = cm->cm_complete_data;
if (cm->cm_data != NULL) {
bus_dmamap_sync(sc->buffer_dmat, cm->cm_dmamap,
BUS_DMASYNC_POSTREAD);
bus_dmamap_unload(sc->buffer_dmat, cm->cm_dmamap);
}
Fix several issues with the mps(4) driver. When the driver ran out of DMA chaining buffers, it kept the timeout for the I/O, and I/O would stall. The driver was not freezing the device queue on errors. mps.c: Pull command completion logic into a separate function, and call the callback/wakeup for commands that are never sent due to lack of chain buffers. Add a number of extra diagnostic sysctl variables. Handle pre-hardware errors for configuration I/O. This doesn't panic the system, but it will fail the configuration I/O and there is no retry mechanism. So the device probe will not succeed. This should be a very uncommon situation, however. mps_sas.c: Freeze the SIM queue when we run out of chain buffers, and unfreeze it when more commands complete. Freeze the device queue when errors occur, so that CAM can insure proper command ordering. Report pre-hardware errors for task management commands. In general, that shouldn't be possible because task management commands don't have S/G lists, and that is currently the only error path before we get to the hardware. Handle pre-hardware errors (like out of chain elements) for SMP requests. That shouldn't happen either, since we should have enough space for two S/G elements in the standard request. For commands that end with MPI2_IOCSTATUS_SCSI_IOC_TERMINATED and MPI2_IOCSTATUS_SCSI_EXT_TERMINATED, return them with CAM_REQUEUE_REQ to retry them unconditionally. These seem to be related to back end, transport related problems that are hopefully transient. We don't want to go through the retry count for something that is not a permanent error. Keep track of the number of outstanding I/Os. mpsvar.h: Track the number of free chain elements. Add variables for the number of outstanding I/Os, and I/O high water mark. Add variables to track the number of free chain buffers and the chain low water mark, as well as the number of chain allocation failures. Add I/O state flags and an attach done flag. MFC after: 3 days
2011-02-18 17:06:06 +00:00
/*
* XXX KDM need to do more error recovery? This results in the
* device in question not getting probed.
*/
if ((cm->cm_flags & MPS_CM_FLAGS_ERROR_MASK) != 0) {
params->status = MPI2_IOCSTATUS_BUSY;
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
goto done;
Fix several issues with the mps(4) driver. When the driver ran out of DMA chaining buffers, it kept the timeout for the I/O, and I/O would stall. The driver was not freezing the device queue on errors. mps.c: Pull command completion logic into a separate function, and call the callback/wakeup for commands that are never sent due to lack of chain buffers. Add a number of extra diagnostic sysctl variables. Handle pre-hardware errors for configuration I/O. This doesn't panic the system, but it will fail the configuration I/O and there is no retry mechanism. So the device probe will not succeed. This should be a very uncommon situation, however. mps_sas.c: Freeze the SIM queue when we run out of chain buffers, and unfreeze it when more commands complete. Freeze the device queue when errors occur, so that CAM can insure proper command ordering. Report pre-hardware errors for task management commands. In general, that shouldn't be possible because task management commands don't have S/G lists, and that is currently the only error path before we get to the hardware. Handle pre-hardware errors (like out of chain elements) for SMP requests. That shouldn't happen either, since we should have enough space for two S/G elements in the standard request. For commands that end with MPI2_IOCSTATUS_SCSI_IOC_TERMINATED and MPI2_IOCSTATUS_SCSI_EXT_TERMINATED, return them with CAM_REQUEUE_REQ to retry them unconditionally. These seem to be related to back end, transport related problems that are hopefully transient. We don't want to go through the retry count for something that is not a permanent error. Keep track of the number of outstanding I/Os. mpsvar.h: Track the number of free chain elements. Add variables for the number of outstanding I/Os, and I/O high water mark. Add variables to track the number of free chain buffers and the chain low water mark, as well as the number of chain allocation failures. Add I/O state flags and an attach done flag. MFC after: 3 days
2011-02-18 17:06:06 +00:00
}
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
reply = (MPI2_CONFIG_REPLY *)cm->cm_reply;
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
if (reply == NULL) {
params->status = MPI2_IOCSTATUS_BUSY;
goto done;
}
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
params->status = reply->IOCStatus;
if (params->hdr.Ext.ExtPageType != 0) {
params->hdr.Ext.ExtPageType = reply->ExtPageType;
params->hdr.Ext.ExtPageLength = reply->ExtPageLength;
} else {
params->hdr.Struct.PageType = reply->Header.PageType;
params->hdr.Struct.PageNumber = reply->Header.PageNumber;
params->hdr.Struct.PageLength = reply->Header.PageLength;
params->hdr.Struct.PageVersion = reply->Header.PageVersion;
}
Bring in the LSI-supported version of the mps(4) driver. This involves significant changes to the mps(4) driver, but is not a complete rewrite. Some of the changes in this version of the driver: - Integrated RAID (IR) support. - Support for WarpDrive controllers. - Support for SCSI protection information (EEDP). - Support for TLR (Transport Level Retries), needed for tape drives. - Improved error recovery code. - ioctl interface compatible with LSI utilities. mps.4: Update the mps(4) driver man page somewhat for the driver changes. The list of supported hardware still needs to be updated to reflect the full list of supported cards. conf/files: Add the new driver files. mps/mpi/*: Updated version of the MPI header files, with a BSD style copyright. mps/*: See above for a description of the new driver features. modules/mps/Makefile: Add the new mps(4) driver files. Submitted by: Kashyap Desai <Kashyap.Desai@lsi.com> Reviewed by: ken MFC after: 1 week
2012-01-26 18:17:21 +00:00
done:
MFp4 (//depot/projects/mps/...) Bring in a driver for the LSI Logic MPT2 6Gb SAS controllers. This driver supports basic I/O, and works with SAS and SATA drives and expanders. Basic error recovery works (i.e. timeouts and aborts) as well. Integrated RAID isn't supported yet, and there are some known bugs. So this isn't ready for production use, but is certainly ready for testing and additional development. For the moment, new commits to this driver should go into the FreeBSD Perforce repository first (//depot/projects/mps/...) and then get merged into -current once they've been vetted. This has only been added to the amd64 GENERIC, since that is the only architecture I have tested this driver with. Submitted by: scottl Discussed with: imp, gibbs, will Sponsored by: Yahoo, Spectra Logic Corporation
2010-09-10 15:03:56 +00:00
mps_free_command(sc, cm);
if (params->callback != NULL)
params->callback(sc, params);
return;
}
Reference in New Issue Copy Permalink