d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
82b42c364a
freebsd-skq/sys/dev/mpt/mpt.h

1177 lines
36 KiB
C
Raw Normal View History

Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
/* $FreeBSD$ */
Start each of the license/copyright comments with /*-, minor shuffle of lines
2005-01-06 01:43:34 +00:00
/*-
Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
* Generic defines for LSI '909 FC adapters.
* FreeBSD Version.
*
* Copyright (c) 2000, 2001 by Greg Ansley
*
* 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 immediately at the beginning of the file, without modification,
* this list of conditions, and the following disclaimer.
* 2. The name of the author may not be used to endorse or promote products
* derived from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
* ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
Role a microrev of the MPI Library in preparation for target mode work. Make my portions of the license clearer. Thank Chris Ellsworth for his support in getting a bunch of this done.
2006-02-25 07:45:54 +00:00
*/
/*-
* Copyright (c) 2002, 2006 by Matthew Jacob
* 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 at minimum a disclaimer
* substantially similar to the "NO WARRANTY" disclaimer below
* ("Disclaimer") and any redistribution must be conditioned upon including
* a substantially similar Disclaimer requirement for further binary
* redistribution.
* 3. Neither the names of the above listed copyright holders nor the names
* of any contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS 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 COPYRIGHT OWNER 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 THE COPYRIGHT
* OWNER OR CONTRIBUTOR IS ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
*
Role a microrev of the MPI Library in preparation for target mode work. Make my portions of the license clearer. Thank Chris Ellsworth for his support in getting a bunch of this done.
2006-02-25 07:45:54 +00:00
* Support from Chris Ellsworth in order to make SAS adapters work
* is gratefully acknowledged.
Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
*/
/*
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
* Copyright (c) 2004, Avid Technology, Inc. and its contributors.
* Copyright (c) 2004, 2005 Justin T. Gibbs
* Copyright (c) 2005, WHEEL Sp. z o.o.
* 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 at minimum a disclaimer
* substantially similar to the "NO WARRANTY" disclaimer below
* ("Disclaimer") and any redistribution must be conditioned upon including
* a substantially similar Disclaimer requirement for further binary
* redistribution.
Correct attribution in clause three to address the correct copyright holders. The license that was approved for my changes to this driver originally came from LSI, but the changes to the driver core are not owned by LSI. MFC: 1 day
2005-08-03 14:08:41 +00:00
* 3. Neither the names of the above listed copyright holders nor the names
* of any contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS 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 COPYRIGHT OWNER 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 THE COPYRIGHT
* OWNER OR CONTRIBUTOR IS ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
*/
#ifndef _MPT_H_
#define _MPT_H_
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
/********************************* OS Includes ********************************/
#include <sys/types.h>
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/endian.h>
#include <sys/eventhandler.h>
#if __FreeBSD_version < 500000
#include <sys/kernel.h>
#include <sys/queue.h>
#include <sys/malloc.h>
#else
#include <sys/lock.h>
#include <sys/kernel.h>
#include <sys/queue.h>
#include <sys/malloc.h>
#include <sys/mutex.h>
#include <sys/condvar.h>
#endif
#include <sys/proc.h>
#include <sys/bus.h>
#include <sys/module.h>
#include <machine/clock.h>
#include <machine/cpu.h>
#include <machine/resource.h>
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
#if __FreeBSD_version < 500000
#include <machine/bus.h>
#endif
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
#include <sys/rman.h>
Do initial cut of SAS HBA support. These controllers (106X) seem to support automatically both SATA and SAS drives. The async SAS event handling we catch but ignore at present (so automagic attach/detach isn't hooked up yet). Do 64 bit PCI support- we can now work on systems with > 4GB of memory. Do large transfer support- we now can support up to reported chain depth, or the length of our request area. We simply allocate additional request elements when we would run out of room for chain lists. Tested on Ultra320, FC and SAS controllers on AMD64 and i386 platforms. There were no RAID cards available for me to regression test. The error recovery for this driver still is pretty bad.
2006-02-11 01:35:29 +00:00
#if __FreeBSD_version < 500000
#include <pci/pcireg.h>
#include <pci/pcivar.h>
#else
#include <dev/pci/pcireg.h>
#include <dev/pci/pcivar.h>
#endif
#include <machine/bus.h>
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
#include "opt_ddb.h"
/**************************** Register Definitions ****************************/
#include <dev/mpt/mpt_reg.h>
/******************************* MPI Definitions ******************************/
#include <dev/mpt/mpilib/mpi_type.h>
#include <dev/mpt/mpilib/mpi.h>
#include <dev/mpt/mpilib/mpi_cnfg.h>
#include <dev/mpt/mpilib/mpi_ioc.h>
#include <dev/mpt/mpilib/mpi_raid.h>
/* XXX For mpt_debug.c */
#include <dev/mpt/mpilib/mpi_init.h>
/****************************** Misc Definitions ******************************/
Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
#define MPT_OK (0)
#define MPT_FAIL (0x10000)
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
#define NUM_ELEMENTS(array) (sizeof(array) / sizeof(*array))
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
#define MPT_ROLE_NONE 0
#define MPT_ROLE_INITIATOR 1
#define MPT_ROLE_TARGET 2
#define MPT_ROLE_BOTH 3
#define MPT_ROLE_DEFAULT MPT_ROLE_INITIATOR
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
/**************************** Forward Declarations ****************************/
struct mpt_softc;
struct mpt_personality;
typedef struct req_entry request_t;
/************************* Personality Module Support *************************/
typedef int mpt_load_handler_t(struct mpt_personality *);
typedef int mpt_probe_handler_t(struct mpt_softc *);
typedef int mpt_attach_handler_t(struct mpt_softc *);
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
typedef int mpt_enable_handler_t(struct mpt_softc *);
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
typedef int mpt_event_handler_t(struct mpt_softc *, request_t *,
MSG_EVENT_NOTIFY_REPLY *);
typedef void mpt_reset_handler_t(struct mpt_softc *, int /*type*/);
/* XXX Add return value and use for veto? */
typedef void mpt_shutdown_handler_t(struct mpt_softc *);
typedef void mpt_detach_handler_t(struct mpt_softc *);
typedef int mpt_unload_handler_t(struct mpt_personality *);
struct mpt_personality
{
const char *name;
uint32_t id; /* Assigned identifier. */
u_int use_count; /* Instances using personality*/
mpt_load_handler_t *load; /* configure personailty */
#define MPT_PERS_FIRST_HANDLER(pers) (&(pers)->load)
mpt_probe_handler_t *probe; /* configure personailty */
mpt_attach_handler_t *attach; /* initialize device instance */
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
mpt_enable_handler_t *enable; /* enable device */
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
mpt_event_handler_t *event; /* Handle MPI event. */
mpt_reset_handler_t *reset; /* Re-init after reset. */
mpt_shutdown_handler_t *shutdown; /* Shutdown instance. */
mpt_detach_handler_t *detach; /* release device instance */
mpt_unload_handler_t *unload; /* Shutdown personality */
#define MPT_PERS_LAST_HANDLER(pers) (&(pers)->unload)
Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
};
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
int mpt_modevent(module_t, int, void *);
/* Maximum supported number of personalities. */
#define MPT_MAX_PERSONALITIES (15)
#define MPT_PERSONALITY_DEPEND(name, dep, vmin, vpref, vmax) \
MODULE_DEPEND(name, dep, vmin, vpref, vmax)
#define DECLARE_MPT_PERSONALITY(name, order) \
static moduledata_t name##_mod = { \
#name, mpt_modevent, &name##_personality \
}; \
DECLARE_MODULE(name, name##_mod, SI_SUB_DRIVERS, order); \
MODULE_VERSION(name, 1); \
MPT_PERSONALITY_DEPEND(name, mpt_core, 1, 1, 1)
/******************************* Bus DMA Support ******************************/
/* XXX Need to update bus_dmamap_sync to take a range argument. */
#define bus_dmamap_sync_range(dma_tag, dmamap, offset, len, op) \
bus_dmamap_sync(dma_tag, dmamap, op)
#if __FreeBSD_version >= 501102
#define mpt_dma_tag_create(mpt, parent_tag, alignment, boundary, \
lowaddr, highaddr, filter, filterarg, \
maxsize, nsegments, maxsegsz, flags, \
dma_tagp) \
bus_dma_tag_create(parent_tag, alignment, boundary, \
lowaddr, highaddr, filter, filterarg, \
maxsize, nsegments, maxsegsz, flags, \
busdma_lock_mutex, &Giant, \
dma_tagp)
#else
#define mpt_dma_tag_create(mpt, parent_tag, alignment, boundary, \
lowaddr, highaddr, filter, filterarg, \
maxsize, nsegments, maxsegsz, flags, \
dma_tagp) \
bus_dma_tag_create(parent_tag, alignment, boundary, \
lowaddr, highaddr, filter, filterarg, \
maxsize, nsegments, maxsegsz, flags, \
dma_tagp)
#endif
struct mpt_map_info {
struct mpt_softc *mpt;
int error;
uint32_t phys;
};
void mpt_map_rquest(void *, bus_dma_segment_t *, int, int);
/**************************** Kernel Thread Support ***************************/
#if __FreeBSD_version > 500005
#define mpt_kthread_create(func, farg, proc_ptr, flags, stackpgs, fmtstr, arg) \
kthread_create(func, farg, proc_ptr, flags, stackpgs, fmtstr, arg)
#else
#define mpt_kthread_create(func, farg, proc_ptr, flags, stackpgs, fmtstr, arg) \
kthread_create(func, farg, proc_ptr, fmtstr, arg)
#endif
/****************************** Timer Facilities ******************************/
#if __FreeBSD_version > 500000
#define mpt_callout_init(c) callout_init(c, /*mpsafe*/0);
#else
#define mpt_callout_init(c) callout_init(c);
#endif
Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
/********************************** Endianess *********************************/
static __inline uint64_t
u64toh(U64 s)
{
uint64_t result;
Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
result = le32toh(s.Low);
result |= ((uint64_t)le32toh(s.High)) << 32;
return (result);
}
Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
/**************************** MPI Transaction State ***************************/
typedef enum {
Fix some of the previus changes 'better'. There's something strange going on with async events. They seem to be be treated differently for different Fusion implementations. Some will really tell you when it's okay to free the request that started them. Some won't. Very disconcerting. This is particularily bad when the chip (FC in this case) tells you in the reply that it's not a continuation reply, which means you can free the request that its associated with. However, if you do that, I've found that additional async event replies come back for that message context after you freed it. Very Bad Things Happen. Put in a reply register debounce. Warn about out of range context indices. Use more MPILIB defines where possible. Replace bzero with memset. Add tons more KASSERTS. Do a *lot* more request free list auditting and serial number usages. Get rid of the warning about the short IOC Facts Reply. Go back to 16 bits of context index. Do a lot more target state auditting as well. Make a tag out of not only the ioindex but the request index as well and worry less about keeping a full serial number.
2006-04-01 07:12:18 +00:00
REQ_STATE_NIL = 0x00,
REQ_STATE_FREE = 0x01,
REQ_STATE_ALLOCATED = 0x02,
REQ_STATE_QUEUED = 0x04,
REQ_STATE_DONE = 0x08,
REQ_STATE_TIMEDOUT = 0x10,
REQ_STATE_NEED_WAKEUP = 0x20,
REQ_STATE_LOCKED = 0x80, /* can't be freed */
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
REQ_STATE_MASK = 0xFF
} mpt_req_state_t;
struct req_entry {
TAILQ_ENTRY(req_entry) links; /* Pointer to next in list */
mpt_req_state_t state; /* Request State Information */
uint16_t index; /* Index of this entry */
uint16_t IOCStatus; /* Completion status */
A large set of changes: + Add boatloads of KASSERTs and *really* check out more locking issues (to catch recursions when we actually go to real locking in CAM soon). The KASSERTs also caught lots of other issues like using commands that were put back on free lists, etc. + Target mode: role setting is derived directly from port capabilities. There is no need to set a role any more. Some target mode resources are allocated early on (ELS), but target command buffer allocation is deferred until the first lun enable. + Fix some breakages I introduced with target mode in that some commands are *repeating* commands. That is, the reply shows up but the command isn't really done (we don't free it). We still need to take it off the pending list because when we resubmit it, bad things then happen. + Fix more of the way that timed out commands and bus reset is done. The actual TMF response code was being ignored. + For SPI, honor BIOS settings. This doesn't quite fix the problems we've seen where we can't seem to (re)negotiate U320 on all drives but avoids it instead by letting us honor the BIOS settings. I'm sure this is not quite right and will have to change again soon.
2006-04-11 16:47:30 +00:00
uint16_t ResponseCode; /* TMF Reponse Code */
uint16_t serno; /* serial number */
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
union ccb *ccb; /* CAM request */
void *req_vbuf; /* Virtual Address of Entry */
void *sense_vbuf; /* Virtual Address of sense data */
bus_addr_t req_pbuf; /* Physical Address of Entry */
bus_addr_t sense_pbuf; /* Physical Address of sense data */
bus_dmamap_t dmap; /* DMA map for data buffer */
Do initial cut of SAS HBA support. These controllers (106X) seem to support automatically both SATA and SAS drives. The async SAS event handling we catch but ignore at present (so automagic attach/detach isn't hooked up yet). Do 64 bit PCI support- we can now work on systems with > 4GB of memory. Do large transfer support- we now can support up to reported chain depth, or the length of our request area. We simply allocate additional request elements when we would run out of room for chain lists. Tested on Ultra320, FC and SAS controllers on AMD64 and i386 platforms. There were no RAID cards available for me to regression test. The error recovery for this driver still is pretty bad.
2006-02-11 01:35:29 +00:00
struct req_entry *chain; /* for SGE overallocations */
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
};
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
/**************************** MPI Target State Info ***************************/
typedef struct {
uint32_t reply_desc; /* current reply descriptor */
uint32_t resid; /* current data residual */
uint32_t bytes_xfered; /* current relative offset */
union ccb *ccb; /* pointer to currently active ccb */
request_t *req; /* pointer to currently active assist request */
int nxfers;
Fix some of the previus changes 'better'. There's something strange going on with async events. They seem to be be treated differently for different Fusion implementations. Some will really tell you when it's okay to free the request that started them. Some won't. Very disconcerting. This is particularily bad when the chip (FC in this case) tells you in the reply that it's not a continuation reply, which means you can free the request that its associated with. However, if you do that, I've found that additional async event replies come back for that message context after you freed it. Very Bad Things Happen. Put in a reply register debounce. Warn about out of range context indices. Use more MPILIB defines where possible. Replace bzero with memset. Add tons more KASSERTS. Do a *lot* more request free list auditting and serial number usages. Get rid of the warning about the short IOC Facts Reply. Go back to 16 bits of context index. Do a lot more target state auditting as well. Make a tag out of not only the ioindex but the request index as well and worry less about keeping a full serial number.
2006-04-01 07:12:18 +00:00
uint32_t tag_id;
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
enum {
TGT_STATE_NIL,
Fix some of the previus changes 'better'. There's something strange going on with async events. They seem to be be treated differently for different Fusion implementations. Some will really tell you when it's okay to free the request that started them. Some won't. Very disconcerting. This is particularily bad when the chip (FC in this case) tells you in the reply that it's not a continuation reply, which means you can free the request that its associated with. However, if you do that, I've found that additional async event replies come back for that message context after you freed it. Very Bad Things Happen. Put in a reply register debounce. Warn about out of range context indices. Use more MPILIB defines where possible. Replace bzero with memset. Add tons more KASSERTS. Do a *lot* more request free list auditting and serial number usages. Get rid of the warning about the short IOC Facts Reply. Go back to 16 bits of context index. Do a lot more target state auditting as well. Make a tag out of not only the ioindex but the request index as well and worry less about keeping a full serial number.
2006-04-01 07:12:18 +00:00
TGT_STATE_LOADING,
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
TGT_STATE_LOADED,
TGT_STATE_IN_CAM,
TGT_STATE_SETTING_UP_FOR_DATA,
TGT_STATE_MOVING_DATA,
TGT_STATE_MOVING_DATA_AND_STATUS,
TGT_STATE_SENDING_STATUS
} state;
} mpt_tgt_state_t;
/*
* When we get an incoming command it has its own tag which is called the
* IoIndex. This is the value we gave that particular command buffer when
* we originally assigned it. It's just a number, really. The FC card uses
* it as an RX_ID. We can use it to index into mpt->tgt_cmd_ptrs, which
* contains pointers the request_t structures related to that IoIndex.
*
* What *we* do is construct a tag out of the index for the target command
* which owns the incoming ATIO plus a rolling sequence number.
*/
#define MPT_MAKE_TAGID(mpt, req, ioindex) \
Fix some of the previus changes 'better'. There's something strange going on with async events. They seem to be be treated differently for different Fusion implementations. Some will really tell you when it's okay to free the request that started them. Some won't. Very disconcerting. This is particularily bad when the chip (FC in this case) tells you in the reply that it's not a continuation reply, which means you can free the request that its associated with. However, if you do that, I've found that additional async event replies come back for that message context after you freed it. Very Bad Things Happen. Put in a reply register debounce. Warn about out of range context indices. Use more MPILIB defines where possible. Replace bzero with memset. Add tons more KASSERTS. Do a *lot* more request free list auditting and serial number usages. Get rid of the warning about the short IOC Facts Reply. Go back to 16 bits of context index. Do a lot more target state auditting as well. Make a tag out of not only the ioindex but the request index as well and worry less about keeping a full serial number.
2006-04-01 07:12:18 +00:00
((ioindex << 18) | (((mpt->sequence++) & 0x3f) << 12) | (req->index & 0xfff))
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
#ifdef INVARIANTS
#define MPT_TAG_2_REQ(a, b) mpt_tag_2_req(a, (uint32_t) b)
#else
Fix some of the previus changes 'better'. There's something strange going on with async events. They seem to be be treated differently for different Fusion implementations. Some will really tell you when it's okay to free the request that started them. Some won't. Very disconcerting. This is particularily bad when the chip (FC in this case) tells you in the reply that it's not a continuation reply, which means you can free the request that its associated with. However, if you do that, I've found that additional async event replies come back for that message context after you freed it. Very Bad Things Happen. Put in a reply register debounce. Warn about out of range context indices. Use more MPILIB defines where possible. Replace bzero with memset. Add tons more KASSERTS. Do a *lot* more request free list auditting and serial number usages. Get rid of the warning about the short IOC Facts Reply. Go back to 16 bits of context index. Do a lot more target state auditting as well. Make a tag out of not only the ioindex but the request index as well and worry less about keeping a full serial number.
2006-04-01 07:12:18 +00:00
#define MPT_TAG_2_REQ(mpt, tag) mpt->tgt_cmd_ptrs[tag >> 18]
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
#endif
#define MPT_TGT_STATE(mpt, req) ((mpt_tgt_state_t *) \
(&((uint8_t *)req->req_vbuf)[MPT_RQSL(mpt) - sizeof (mpt_tgt_state_t)]))
STAILQ_HEAD(mpt_hdr_stailq, ccb_hdr);
#define MPT_MAX_LUNS 256
typedef struct {
struct mpt_hdr_stailq atios;
struct mpt_hdr_stailq inots;
int enabled;
} tgt_resource_t;
A large set of changes: + Add boatloads of KASSERTs and *really* check out more locking issues (to catch recursions when we actually go to real locking in CAM soon). The KASSERTs also caught lots of other issues like using commands that were put back on free lists, etc. + Target mode: role setting is derived directly from port capabilities. There is no need to set a role any more. Some target mode resources are allocated early on (ELS), but target command buffer allocation is deferred until the first lun enable. + Fix some breakages I introduced with target mode in that some commands are *repeating* commands. That is, the reply shows up but the command isn't really done (we don't free it). We still need to take it off the pending list because when we resubmit it, bad things then happen. + Fix more of the way that timed out commands and bus reset is done. The actual TMF response code was being ignored. + For SPI, honor BIOS settings. This doesn't quite fix the problems we've seen where we can't seem to (re)negotiate U320 on all drives but avoids it instead by letting us honor the BIOS settings. I'm sure this is not quite right and will have to change again soon.
2006-04-11 16:47:30 +00:00
#define MPT_MAX_ELS 64
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
/**************************** Handler Registration ****************************/
Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
/*
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
* Global table of registered reply handlers. The
* handler is indicated by byte 3 of the request
* index submitted to the IOC. This allows the
* driver core to perform generic processing without
* any knowledge of per-personality behavior.
*
* MPT_NUM_REPLY_HANDLERS must be a power of 2
* to allow the easy generation of a mask.
*
* The handler offsets used by the core are hard coded
* allowing faster code generation when assigning a handler
* to a request. All "personalities" must use the
* the handler registration mechanism.
*
* The IOC handlers that are rarely executed are placed
* at the tail of the table to make it more likely that
* all commonly executed handlers fit in a single cache
* line.
Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
*/
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
#define MPT_NUM_REPLY_HANDLERS (32)
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
#define MPT_REPLY_HANDLER_EVENTS MPT_CBI_TO_HID(0)
#define MPT_REPLY_HANDLER_CONFIG MPT_CBI_TO_HID(MPT_NUM_REPLY_HANDLERS-1)
#define MPT_REPLY_HANDLER_HANDSHAKE MPT_CBI_TO_HID(MPT_NUM_REPLY_HANDLERS-2)
typedef int mpt_reply_handler_t(struct mpt_softc *mpt, request_t *request,
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
uint32_t reply_desc, MSG_DEFAULT_REPLY *reply_frame);
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
typedef union {
mpt_reply_handler_t *reply_handler;
} mpt_handler_t;
typedef enum {
MPT_HANDLER_REPLY,
MPT_HANDLER_EVENT,
MPT_HANDLER_RESET,
MPT_HANDLER_SHUTDOWN
} mpt_handler_type;
struct mpt_handler_record
{
LIST_ENTRY(mpt_handler_record) links;
mpt_handler_t handler;
};
LIST_HEAD(mpt_handler_list, mpt_handler_record);
/*
* The handler_id is currently unused but would contain the
* handler ID used in the MsgContext field to allow direction
* of replies to the handler. Registrations that don't require
* a handler id can pass in NULL for the handler_id.
*
* Deregistrations for handlers without a handler id should
* pass in MPT_HANDLER_ID_NONE.
*/
#define MPT_HANDLER_ID_NONE (0xFFFFFFFF)
int mpt_register_handler(struct mpt_softc *, mpt_handler_type,
mpt_handler_t, uint32_t *);
int mpt_deregister_handler(struct mpt_softc *, mpt_handler_type,
mpt_handler_t, uint32_t);
/******************* Per-Controller Instance Data Structures ******************/
TAILQ_HEAD(req_queue, req_entry);
/* Structure for saving proper values for modifyable PCI config registers */
struct mpt_pci_cfg {
uint16_t Command;
uint16_t LatencyTimer_LineSize;
uint32_t IO_BAR;
uint32_t Mem0_BAR[2];
uint32_t Mem1_BAR[2];
uint32_t ROM_BAR;
uint8_t IntLine;
uint32_t PMCSR;
Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
};
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
typedef enum {
MPT_RVF_NONE = 0x0,
MPT_RVF_ACTIVE = 0x1,
MPT_RVF_ANNOUNCED = 0x2,
MPT_RVF_UP2DATE = 0x4,
MPT_RVF_REFERENCED = 0x8,
MPT_RVF_WCE_CHANGED = 0x10
} mpt_raid_volume_flags;
struct mpt_raid_volume {
CONFIG_PAGE_RAID_VOL_0 *config_page;
MPI_RAID_VOL_INDICATOR sync_progress;
mpt_raid_volume_flags flags;
u_int quieced_disks;
Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
};
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
typedef enum {
MPT_RDF_NONE = 0x00,
MPT_RDF_ACTIVE = 0x01,
MPT_RDF_ANNOUNCED = 0x02,
MPT_RDF_UP2DATE = 0x04,
MPT_RDF_REFERENCED = 0x08,
MPT_RDF_QUIESCING = 0x10,
MPT_RDF_QUIESCED = 0x20
} mpt_raid_disk_flags;
Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
struct mpt_raid_disk {
CONFIG_PAGE_RAID_PHYS_DISK_0 config_page;
struct mpt_raid_volume *volume;
u_int member_number;
u_int pass_thru_active;
mpt_raid_disk_flags flags;
Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
};
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
struct mpt_evtf_record {
MSG_EVENT_NOTIFY_REPLY reply;
uint32_t context;
LIST_ENTRY(mpt_evtf_record) links;
Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
};
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
LIST_HEAD(mpt_evtf_list, mpt_evtf_record);
struct mpt_softc {
device_t dev;
#if __FreeBSD_version < 500000
uint32_t mpt_islocked;
A large set of changes: + Add boatloads of KASSERTs and *really* check out more locking issues (to catch recursions when we actually go to real locking in CAM soon). The KASSERTs also caught lots of other issues like using commands that were put back on free lists, etc. + Target mode: role setting is derived directly from port capabilities. There is no need to set a role any more. Some target mode resources are allocated early on (ELS), but target command buffer allocation is deferred until the first lun enable. + Fix some breakages I introduced with target mode in that some commands are *repeating* commands. That is, the reply shows up but the command isn't really done (we don't free it). We still need to take it off the pending list because when we resubmit it, bad things then happen. + Fix more of the way that timed out commands and bus reset is done. The actual TMF response code was being ignored. + For SPI, honor BIOS settings. This doesn't quite fix the problems we've seen where we can't seem to (re)negotiate U320 on all drives but avoids it instead by letting us honor the BIOS settings. I'm sure this is not quite right and will have to change again soon.
2006-04-11 16:47:30 +00:00
int mpt_splsaved;
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
#else
struct mtx mpt_lock;
A large set of changes: + Add boatloads of KASSERTs and *really* check out more locking issues (to catch recursions when we actually go to real locking in CAM soon). The KASSERTs also caught lots of other issues like using commands that were put back on free lists, etc. + Target mode: role setting is derived directly from port capabilities. There is no need to set a role any more. Some target mode resources are allocated early on (ELS), but target command buffer allocation is deferred until the first lun enable. + Fix some breakages I introduced with target mode in that some commands are *repeating* commands. That is, the reply shows up but the command isn't really done (we don't free it). We still need to take it off the pending list because when we resubmit it, bad things then happen. + Fix more of the way that timed out commands and bus reset is done. The actual TMF response code was being ignored. + For SPI, honor BIOS settings. This doesn't quite fix the problems we've seen where we can't seem to (re)negotiate U320 on all drives but avoids it instead by letting us honor the BIOS settings. I'm sure this is not quite right and will have to change again soon.
2006-04-11 16:47:30 +00:00
int mpt_locksetup;
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
#endif
uint32_t mpt_pers_mask;
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
uint32_t : 8,
unit : 8,
: 1,
twildcard : 1,
tenabled : 1,
A large set of changes: + Add boatloads of KASSERTs and *really* check out more locking issues (to catch recursions when we actually go to real locking in CAM soon). The KASSERTs also caught lots of other issues like using commands that were put back on free lists, etc. + Target mode: role setting is derived directly from port capabilities. There is no need to set a role any more. Some target mode resources are allocated early on (ELS), but target command buffer allocation is deferred until the first lun enable. + Fix some breakages I introduced with target mode in that some commands are *repeating* commands. That is, the reply shows up but the command isn't really done (we don't free it). We still need to take it off the pending list because when we resubmit it, bad things then happen. + Fix more of the way that timed out commands and bus reset is done. The actual TMF response code was being ignored. + For SPI, honor BIOS settings. This doesn't quite fix the problems we've seen where we can't seem to (re)negotiate U320 on all drives but avoids it instead by letting us honor the BIOS settings. I'm sure this is not quite right and will have to change again soon.
2006-04-11 16:47:30 +00:00
: 2,
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
role : 2, /* none, ini, target, both */
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
raid_mwce_set : 1,
getreqwaiter : 1,
shutdwn_raid : 1,
shutdwn_recovery: 1,
outofbeer : 1,
A large set of changes: + Add boatloads of KASSERTs and *really* check out more locking issues (to catch recursions when we actually go to real locking in CAM soon). The KASSERTs also caught lots of other issues like using commands that were put back on free lists, etc. + Target mode: role setting is derived directly from port capabilities. There is no need to set a role any more. Some target mode resources are allocated early on (ELS), but target command buffer allocation is deferred until the first lun enable. + Fix some breakages I introduced with target mode in that some commands are *repeating* commands. That is, the reply shows up but the command isn't really done (we don't free it). We still need to take it off the pending list because when we resubmit it, bad things then happen. + Fix more of the way that timed out commands and bus reset is done. The actual TMF response code was being ignored. + For SPI, honor BIOS settings. This doesn't quite fix the problems we've seen where we can't seem to (re)negotiate U320 on all drives but avoids it instead by letting us honor the BIOS settings. I'm sure this is not quite right and will have to change again soon.
2006-04-11 16:47:30 +00:00
: 1,
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
disabled : 1,
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
is_sas : 1,
is_fc : 1;
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
u_int verbose;
/*
* IOC Facts
*/
uint16_t mpt_global_credits;
uint16_t request_frame_size;
uint8_t mpt_max_devices;
uint8_t mpt_max_buses;
Do initial cut of SAS HBA support. These controllers (106X) seem to support automatically both SATA and SAS drives. The async SAS event handling we catch but ignore at present (so automagic attach/detach isn't hooked up yet). Do 64 bit PCI support- we can now work on systems with > 4GB of memory. Do large transfer support- we now can support up to reported chain depth, or the length of our request area. We simply allocate additional request elements when we would run out of room for chain lists. Tested on Ultra320, FC and SAS controllers on AMD64 and i386 platforms. There were no RAID cards available for me to regression test. The error recovery for this driver still is pretty bad.
2006-02-11 01:35:29 +00:00
uint8_t ioc_facts_flags;
Fix some of the previus changes 'better'. There's something strange going on with async events. They seem to be be treated differently for different Fusion implementations. Some will really tell you when it's okay to free the request that started them. Some won't. Very disconcerting. This is particularily bad when the chip (FC in this case) tells you in the reply that it's not a continuation reply, which means you can free the request that its associated with. However, if you do that, I've found that additional async event replies come back for that message context after you freed it. Very Bad Things Happen. Put in a reply register debounce. Warn about out of range context indices. Use more MPILIB defines where possible. Replace bzero with memset. Add tons more KASSERTS. Do a *lot* more request free list auditting and serial number usages. Get rid of the warning about the short IOC Facts Reply. Go back to 16 bits of context index. Do a lot more target state auditting as well. Make a tag out of not only the ioindex but the request index as well and worry less about keeping a full serial number.
2006-04-01 07:12:18 +00:00
uint8_t padding0;
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
/*
* Port Facts
* XXX - Add multi-port support!.
*/
uint16_t mpt_ini_id;
uint16_t mpt_port_type;
uint16_t mpt_proto_flags;
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
uint16_t mpt_max_tgtcmds;
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
/*
* Device Configuration Information
*/
union {
struct mpt_spi_cfg {
CONFIG_PAGE_SCSI_PORT_0 _port_page0;
CONFIG_PAGE_SCSI_PORT_1 _port_page1;
CONFIG_PAGE_SCSI_PORT_2 _port_page2;
CONFIG_PAGE_SCSI_DEVICE_0 _dev_page0[16];
CONFIG_PAGE_SCSI_DEVICE_1 _dev_page1[16];
uint16_t _tag_enable;
uint16_t _disc_enable;
} spi;
#define mpt_port_page0 cfg.spi._port_page0
#define mpt_port_page1 cfg.spi._port_page1
#define mpt_port_page2 cfg.spi._port_page2
#define mpt_dev_page0 cfg.spi._dev_page0
#define mpt_dev_page1 cfg.spi._dev_page1
#define mpt_tag_enable cfg.spi._tag_enable
#define mpt_disc_enable cfg.spi._disc_enable
struct mpi_fc_cfg {
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
CONFIG_PAGE_FC_PORT_0 _port_page0;
#define mpt_fcport_page0 cfg.fc._port_page0
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
} fc;
} cfg;
/* Controller Info */
CONFIG_PAGE_IOC_2 * ioc_page2;
CONFIG_PAGE_IOC_3 * ioc_page3;
/* Raid Data */
struct mpt_raid_volume* raid_volumes;
struct mpt_raid_disk* raid_disks;
u_int raid_max_volumes;
u_int raid_max_disks;
u_int raid_page0_len;
u_int raid_wakeup;
u_int raid_rescan;
u_int raid_resync_rate;
u_int raid_mwce_setting;
u_int raid_queue_depth;
Keep track of volumes in non-optimal state and expose a simple count of volumes that might need administrator attention through device specific sysctl to simplify device monitoring. Submitted by: Deomid Ryabkov <myself at rojer dot pp dot ru>
2005-11-15 02:38:35 +00:00
u_int raid_nonopt_volumes;
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
struct proc *raid_thread;
struct callout raid_timer;
/*
* PCI Hardware info
*/
struct resource * pci_irq; /* Interrupt map for chip */
void * ih; /* Interupt handle */
struct mpt_pci_cfg pci_cfg; /* saved PCI conf registers */
/*
* DMA Mapping Stuff
*/
struct resource * pci_reg; /* Register map for chip */
int pci_mem_rid; /* Resource ID */
bus_space_tag_t pci_st; /* Bus tag for registers */
bus_space_handle_t pci_sh; /* Bus handle for registers */
/* PIO versions of above. */
int pci_pio_rid;
struct resource * pci_pio_reg;
bus_space_tag_t pci_pio_st;
bus_space_handle_t pci_pio_sh;
bus_dma_tag_t parent_dmat; /* DMA tag for parent PCI bus */
bus_dma_tag_t reply_dmat; /* DMA tag for reply memory */
bus_dmamap_t reply_dmap; /* DMA map for reply memory */
uint8_t *reply; /* KVA of reply memory */
bus_addr_t reply_phys; /* BusAddr of reply memory */
bus_dma_tag_t buffer_dmat; /* DMA tag for buffers */
bus_dma_tag_t request_dmat; /* DMA tag for request memroy */
bus_dmamap_t request_dmap; /* DMA map for request memroy */
uint8_t *request; /* KVA of Request memory */
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
bus_addr_t request_phys; /* BusAddr of request memory */
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
Do initial cut of SAS HBA support. These controllers (106X) seem to support automatically both SATA and SAS drives. The async SAS event handling we catch but ignore at present (so automagic attach/detach isn't hooked up yet). Do 64 bit PCI support- we can now work on systems with > 4GB of memory. Do large transfer support- we now can support up to reported chain depth, or the length of our request area. We simply allocate additional request elements when we would run out of room for chain lists. Tested on Ultra320, FC and SAS controllers on AMD64 and i386 platforms. There were no RAID cards available for me to regression test. The error recovery for this driver still is pretty bad.
2006-02-11 01:35:29 +00:00
uint32_t max_seg_cnt; /* calculated after IOC facts */
/*
* Hardware management
*/
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
u_int reset_cnt;
/*
* CAM && Software Management
*/
request_t *request_pool;
struct req_queue request_free_list;
struct req_queue request_pending_list;
struct req_queue request_timeout_list;
struct cam_sim *sim;
struct cam_path *path;
struct cam_sim *phydisk_sim;
struct cam_path *phydisk_path;
struct proc *recovery_thread;
request_t *tmf_req;
Fix some of the previus changes 'better'. There's something strange going on with async events. They seem to be be treated differently for different Fusion implementations. Some will really tell you when it's okay to free the request that started them. Some won't. Very disconcerting. This is particularily bad when the chip (FC in this case) tells you in the reply that it's not a continuation reply, which means you can free the request that its associated with. However, if you do that, I've found that additional async event replies come back for that message context after you freed it. Very Bad Things Happen. Put in a reply register debounce. Warn about out of range context indices. Use more MPILIB defines where possible. Replace bzero with memset. Add tons more KASSERTS. Do a *lot* more request free list auditting and serial number usages. Get rid of the warning about the short IOC Facts Reply. Go back to 16 bits of context index. Do a lot more target state auditting as well. Make a tag out of not only the ioindex but the request index as well and worry less about keeping a full serial number.
2006-04-01 07:12:18 +00:00
/*
* Deferred frame acks due to resource shortage.
*/
struct mpt_evtf_list ack_frames;
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
/*
* Target Mode Support
*/
uint32_t scsi_tgt_handler_id;
request_t ** tgt_cmd_ptrs;
A large set of changes: + Add boatloads of KASSERTs and *really* check out more locking issues (to catch recursions when we actually go to real locking in CAM soon). The KASSERTs also caught lots of other issues like using commands that were put back on free lists, etc. + Target mode: role setting is derived directly from port capabilities. There is no need to set a role any more. Some target mode resources are allocated early on (ELS), but target command buffer allocation is deferred until the first lun enable. + Fix some breakages I introduced with target mode in that some commands are *repeating* commands. That is, the reply shows up but the command isn't really done (we don't free it). We still need to take it off the pending list because when we resubmit it, bad things then happen. + Fix more of the way that timed out commands and bus reset is done. The actual TMF response code was being ignored. + For SPI, honor BIOS settings. This doesn't quite fix the problems we've seen where we can't seem to (re)negotiate U320 on all drives but avoids it instead by letting us honor the BIOS settings. I'm sure this is not quite right and will have to change again soon.
2006-04-11 16:47:30 +00:00
request_t ** els_cmd_ptrs; /* FC only */
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
/*
* *snork*- this is chosen to be here *just in case* somebody
* forgets to point to it exactly and we index off of trt with
* CAM_LUN_WILDCARD.
*/
tgt_resource_t trt_wildcard; /* wildcard luns */
tgt_resource_t trt[MPT_MAX_LUNS];
uint16_t tgt_cmds_allocated;
A large set of changes: + Add boatloads of KASSERTs and *really* check out more locking issues (to catch recursions when we actually go to real locking in CAM soon). The KASSERTs also caught lots of other issues like using commands that were put back on free lists, etc. + Target mode: role setting is derived directly from port capabilities. There is no need to set a role any more. Some target mode resources are allocated early on (ELS), but target command buffer allocation is deferred until the first lun enable. + Fix some breakages I introduced with target mode in that some commands are *repeating* commands. That is, the reply shows up but the command isn't really done (we don't free it). We still need to take it off the pending list because when we resubmit it, bad things then happen. + Fix more of the way that timed out commands and bus reset is done. The actual TMF response code was being ignored. + For SPI, honor BIOS settings. This doesn't quite fix the problems we've seen where we can't seem to (re)negotiate U320 on all drives but avoids it instead by letting us honor the BIOS settings. I'm sure this is not quite right and will have to change again soon.
2006-04-11 16:47:30 +00:00
uint16_t els_cmds_allocated; /* FC only */
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
uint16_t timeouts; /* timeout count */
uint16_t success; /* successes afer timeout */
A large set of changes: + Add boatloads of KASSERTs and *really* check out more locking issues (to catch recursions when we actually go to real locking in CAM soon). The KASSERTs also caught lots of other issues like using commands that were put back on free lists, etc. + Target mode: role setting is derived directly from port capabilities. There is no need to set a role any more. Some target mode resources are allocated early on (ELS), but target command buffer allocation is deferred until the first lun enable. + Fix some breakages I introduced with target mode in that some commands are *repeating* commands. That is, the reply shows up but the command isn't really done (we don't free it). We still need to take it off the pending list because when we resubmit it, bad things then happen. + Fix more of the way that timed out commands and bus reset is done. The actual TMF response code was being ignored. + For SPI, honor BIOS settings. This doesn't quite fix the problems we've seen where we can't seem to (re)negotiate U320 on all drives but avoids it instead by letting us honor the BIOS settings. I'm sure this is not quite right and will have to change again soon.
2006-04-11 16:47:30 +00:00
uint16_t sequence; /* Sequence Number */
uint16_t pad3;
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
/* Opposing port in a 929 or 1030, or NULL */
struct mpt_softc * mpt2;
/* FW Image management */
uint32_t fw_image_size;
uint8_t *fw_image;
bus_dma_tag_t fw_dmat; /* DMA tag for firmware image */
bus_dmamap_t fw_dmap; /* DMA map for firmware image */
bus_addr_t fw_phys; /* BusAddr of request memory */
/* Shutdown Event Handler. */
eventhandler_tag eh;
TAILQ_ENTRY(mpt_softc) links;
};
A large set of changes: + Add boatloads of KASSERTs and *really* check out more locking issues (to catch recursions when we actually go to real locking in CAM soon). The KASSERTs also caught lots of other issues like using commands that were put back on free lists, etc. + Target mode: role setting is derived directly from port capabilities. There is no need to set a role any more. Some target mode resources are allocated early on (ELS), but target command buffer allocation is deferred until the first lun enable. + Fix some breakages I introduced with target mode in that some commands are *repeating* commands. That is, the reply shows up but the command isn't really done (we don't free it). We still need to take it off the pending list because when we resubmit it, bad things then happen. + Fix more of the way that timed out commands and bus reset is done. The actual TMF response code was being ignored. + For SPI, honor BIOS settings. This doesn't quite fix the problems we've seen where we can't seem to (re)negotiate U320 on all drives but avoids it instead by letting us honor the BIOS settings. I'm sure this is not quite right and will have to change again soon.
2006-04-11 16:47:30 +00:00
static __inline void mpt_assign_serno(struct mpt_softc *, request_t *);
static __inline void
mpt_assign_serno(struct mpt_softc *mpt, request_t *req)
{
if ((req->serno = mpt->sequence++) == 0) {
req->serno = mpt->sequence++;
}
}
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
/***************************** Locking Primitives *****************************/
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
#if __FreeBSD_version < 500000
#define MPT_IFLAGS INTR_TYPE_CAM
#define MPT_LOCK(mpt) mpt_lockspl(mpt)
#define MPT_UNLOCK(mpt) mpt_unlockspl(mpt)
A large set of changes: + Add boatloads of KASSERTs and *really* check out more locking issues (to catch recursions when we actually go to real locking in CAM soon). The KASSERTs also caught lots of other issues like using commands that were put back on free lists, etc. + Target mode: role setting is derived directly from port capabilities. There is no need to set a role any more. Some target mode resources are allocated early on (ELS), but target command buffer allocation is deferred until the first lun enable. + Fix some breakages I introduced with target mode in that some commands are *repeating* commands. That is, the reply shows up but the command isn't really done (we don't free it). We still need to take it off the pending list because when we resubmit it, bad things then happen. + Fix more of the way that timed out commands and bus reset is done. The actual TMF response code was being ignored. + For SPI, honor BIOS settings. This doesn't quite fix the problems we've seen where we can't seem to (re)negotiate U320 on all drives but avoids it instead by letting us honor the BIOS settings. I'm sure this is not quite right and will have to change again soon.
2006-04-11 16:47:30 +00:00
#define MPT_OWNED(mpt) mpt->mpt_islocked
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
#define MPTLOCK_2_CAMLOCK MPT_UNLOCK
#define CAMLOCK_2_MPTLOCK MPT_LOCK
#define MPT_LOCK_SETUP(mpt)
#define MPT_LOCK_DESTROY(mpt)
static __inline void mpt_lockspl(struct mpt_softc *mpt);
static __inline void mpt_unlockspl(struct mpt_softc *mpt);
static __inline void
mpt_lockspl(struct mpt_softc *mpt)
{
int s;
s = splcam();
if (mpt->mpt_islocked++ == 0) {
mpt->mpt_splsaved = s;
} else {
splx(s);
panic("Recursed lock with mask: 0x%x\n", s);
}
}
static __inline void
mpt_unlockspl(struct mpt_softc *mpt)
{
if (mpt->mpt_islocked) {
if (--mpt->mpt_islocked == 0) {
splx(mpt->mpt_splsaved);
}
} else
panic("Negative lock count\n");
}
static __inline int
mpt_sleep(struct mpt_softc *mpt, void *ident, int priority,
const char *wmesg, int timo)
{
int saved_cnt;
int saved_spl;
int error;
KASSERT(mpt->mpt_islocked <= 1, ("Invalid lock count on tsleep"));
saved_cnt = mpt->mpt_islocked;
saved_spl = mpt->mpt_splsaved;
mpt->mpt_islocked = 0;
error = tsleep(ident, priority, wmesg, timo);
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
KASSERT(mpt->mpt_islocked == 0, ("Invalid lock count on wakeup"));
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
mpt->mpt_islocked = saved_cnt;
mpt->mpt_splsaved = saved_spl;
return (error);
}
#else
Fix -Wundef.
2005-12-04 02:12:43 +00:00
#ifdef LOCKING_WORKED_AS_IT_SHOULD
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
#error "Shouldn't Be Here!"
#define MPT_IFLAGS INTR_TYPE_CAM | INTR_ENTROPY | INTR_MPSAFE
#define MPT_LOCK_SETUP(mpt) \
mtx_init(&mpt->mpt_lock, "mpt", NULL, MTX_DEF); \
mpt->mpt_locksetup = 1
#define MPT_LOCK_DESTROY(mpt) \
if (mpt->mpt_locksetup) { \
mtx_destroy(&mpt->mpt_lock); \
mpt->mpt_locksetup = 0; \
}
#define MPT_LOCK(mpt) mtx_lock(&(mpt)->mpt_lock)
#define MPT_UNLOCK(mpt) mtx_unlock(&(mpt)->mpt_lock)
A large set of changes: + Add boatloads of KASSERTs and *really* check out more locking issues (to catch recursions when we actually go to real locking in CAM soon). The KASSERTs also caught lots of other issues like using commands that were put back on free lists, etc. + Target mode: role setting is derived directly from port capabilities. There is no need to set a role any more. Some target mode resources are allocated early on (ELS), but target command buffer allocation is deferred until the first lun enable. + Fix some breakages I introduced with target mode in that some commands are *repeating* commands. That is, the reply shows up but the command isn't really done (we don't free it). We still need to take it off the pending list because when we resubmit it, bad things then happen. + Fix more of the way that timed out commands and bus reset is done. The actual TMF response code was being ignored. + For SPI, honor BIOS settings. This doesn't quite fix the problems we've seen where we can't seem to (re)negotiate U320 on all drives but avoids it instead by letting us honor the BIOS settings. I'm sure this is not quite right and will have to change again soon.
2006-04-11 16:47:30 +00:00
#define MPT_OWNED(mpt) mtx_owned(&(mpt)->mpt_lock)
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
#define MPTLOCK_2_CAMLOCK(mpt) \
mtx_unlock(&(mpt)->mpt_lock); mtx_lock(&Giant)
#define CAMLOCK_2_MPTLOCK(mpt) \
mtx_unlock(&Giant); mtx_lock(&(mpt)->mpt_lock)
#define mpt_sleep(mpt, ident, priority, wmesg, timo) \
msleep(ident, &(mpt)->mpt_lock, priority, wmesg, timo)
A large set of changes: + Add boatloads of KASSERTs and *really* check out more locking issues (to catch recursions when we actually go to real locking in CAM soon). The KASSERTs also caught lots of other issues like using commands that were put back on free lists, etc. + Target mode: role setting is derived directly from port capabilities. There is no need to set a role any more. Some target mode resources are allocated early on (ELS), but target command buffer allocation is deferred until the first lun enable. + Fix some breakages I introduced with target mode in that some commands are *repeating* commands. That is, the reply shows up but the command isn't really done (we don't free it). We still need to take it off the pending list because when we resubmit it, bad things then happen. + Fix more of the way that timed out commands and bus reset is done. The actual TMF response code was being ignored. + For SPI, honor BIOS settings. This doesn't quite fix the problems we've seen where we can't seem to (re)negotiate U320 on all drives but avoids it instead by letting us honor the BIOS settings. I'm sure this is not quite right and will have to change again soon.
2006-04-11 16:47:30 +00:00
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
#else
A large set of changes: + Add boatloads of KASSERTs and *really* check out more locking issues (to catch recursions when we actually go to real locking in CAM soon). The KASSERTs also caught lots of other issues like using commands that were put back on free lists, etc. + Target mode: role setting is derived directly from port capabilities. There is no need to set a role any more. Some target mode resources are allocated early on (ELS), but target command buffer allocation is deferred until the first lun enable. + Fix some breakages I introduced with target mode in that some commands are *repeating* commands. That is, the reply shows up but the command isn't really done (we don't free it). We still need to take it off the pending list because when we resubmit it, bad things then happen. + Fix more of the way that timed out commands and bus reset is done. The actual TMF response code was being ignored. + For SPI, honor BIOS settings. This doesn't quite fix the problems we've seen where we can't seem to (re)negotiate U320 on all drives but avoids it instead by letting us honor the BIOS settings. I'm sure this is not quite right and will have to change again soon.
2006-04-11 16:47:30 +00:00
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
#define MPT_IFLAGS INTR_TYPE_CAM | INTR_ENTROPY
#define MPT_LOCK_SETUP(mpt) do { } while (0)
#define MPT_LOCK_DESTROY(mpt) do { } while (0)
A large set of changes: + Add boatloads of KASSERTs and *really* check out more locking issues (to catch recursions when we actually go to real locking in CAM soon). The KASSERTs also caught lots of other issues like using commands that were put back on free lists, etc. + Target mode: role setting is derived directly from port capabilities. There is no need to set a role any more. Some target mode resources are allocated early on (ELS), but target command buffer allocation is deferred until the first lun enable. + Fix some breakages I introduced with target mode in that some commands are *repeating* commands. That is, the reply shows up but the command isn't really done (we don't free it). We still need to take it off the pending list because when we resubmit it, bad things then happen. + Fix more of the way that timed out commands and bus reset is done. The actual TMF response code was being ignored. + For SPI, honor BIOS settings. This doesn't quite fix the problems we've seen where we can't seem to (re)negotiate U320 on all drives but avoids it instead by letting us honor the BIOS settings. I'm sure this is not quite right and will have to change again soon.
2006-04-11 16:47:30 +00:00
#if 0
#define MPT_LOCK(mpt) \
device_printf(mpt->dev, "LOCK %s:%d\n", __FILE__, __LINE__); \
KASSERT(mpt->mpt_locksetup == 0, \
("recursive lock acquire at %s:%d", __FILE__, __LINE__)); \
mpt->mpt_locksetup = 1
#define MPT_UNLOCK(mpt) \
device_printf(mpt->dev, "UNLK %s:%d\n", __FILE__, __LINE__); \
KASSERT(mpt->mpt_locksetup == 1, \
("release unowned lock at %s:%d", __FILE__, __LINE__)); \
mpt->mpt_locksetup = 0
#else
#define MPT_LOCK(mpt) \
KASSERT(mpt->mpt_locksetup == 0, \
("recursive lock acquire at %s:%d", __FILE__, __LINE__)); \
mpt->mpt_locksetup = 1
#define MPT_UNLOCK(mpt) \
KASSERT(mpt->mpt_locksetup == 1, \
("release unowned lock at %s:%d", __FILE__, __LINE__)); \
mpt->mpt_locksetup = 0
#endif
#define MPT_OWNED(mpt) mpt->mpt_locksetup
#define MPTLOCK_2_CAMLOCK(mpt) MPT_UNLOCK(mpt)
#define CAMLOCK_2_MPTLOCK(mpt) MPT_LOCK(mpt)
static __inline int
mpt_sleep(struct mpt_softc *, void *, int, const char *, int);
static __inline int
mpt_sleep(struct mpt_softc *mpt, void *i, int p, const char *w, int t)
{
int r;
MPT_UNLOCK(mpt);
r = tsleep(i, p, w, t);
MPT_LOCK(mpt);
return (r);
}
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
#endif
#endif
/******************************* Register Access ******************************/
static __inline void mpt_write(struct mpt_softc *, size_t, uint32_t);
static __inline uint32_t mpt_read(struct mpt_softc *, int);
static __inline void mpt_pio_write(struct mpt_softc *, size_t, uint32_t);
static __inline uint32_t mpt_pio_read(struct mpt_softc *, int);
static __inline void
mpt_write(struct mpt_softc *mpt, size_t offset, uint32_t val)
{
bus_space_write_4(mpt->pci_st, mpt->pci_sh, offset, val);
}
static __inline uint32_t
mpt_read(struct mpt_softc *mpt, int offset)
{
return (bus_space_read_4(mpt->pci_st, mpt->pci_sh, offset));
}
/*
* Some operations (e.g. diagnostic register writes while the ARM proccessor
* is disabled), must be performed using "PCI pio" operations. On non-PCI
* busses, these operations likely map to normal register accesses.
*/
static __inline void
mpt_pio_write(struct mpt_softc *mpt, size_t offset, uint32_t val)
{
bus_space_write_4(mpt->pci_pio_st, mpt->pci_pio_sh, offset, val);
}
static __inline uint32_t
mpt_pio_read(struct mpt_softc *mpt, int offset)
{
return (bus_space_read_4(mpt->pci_pio_st, mpt->pci_pio_sh, offset));
}
/*********************** Reply Frame/Request Management ***********************/
/* Max MPT Reply we are willing to accept (must be power of 2) */
Do initial cut of SAS HBA support. These controllers (106X) seem to support automatically both SATA and SAS drives. The async SAS event handling we catch but ignore at present (so automagic attach/detach isn't hooked up yet). Do 64 bit PCI support- we can now work on systems with > 4GB of memory. Do large transfer support- we now can support up to reported chain depth, or the length of our request area. We simply allocate additional request elements when we would run out of room for chain lists. Tested on Ultra320, FC and SAS controllers on AMD64 and i386 platforms. There were no RAID cards available for me to regression test. The error recovery for this driver still is pretty bad.
2006-02-11 01:35:29 +00:00
#define MPT_REPLY_SIZE 256
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
/*
* Must be less than 16384 in order for target mode to work
*/
Do initial cut of SAS HBA support. These controllers (106X) seem to support automatically both SATA and SAS drives. The async SAS event handling we catch but ignore at present (so automagic attach/detach isn't hooked up yet). Do 64 bit PCI support- we can now work on systems with > 4GB of memory. Do large transfer support- we now can support up to reported chain depth, or the length of our request area. We simply allocate additional request elements when we would run out of room for chain lists. Tested on Ultra320, FC and SAS controllers on AMD64 and i386 platforms. There were no RAID cards available for me to regression test. The error recovery for this driver still is pretty bad.
2006-02-11 01:35:29 +00:00
#define MPT_MAX_REQUESTS(mpt) 512
#define MPT_REQUEST_AREA 512
#define MPT_SENSE_SIZE 32 /* included in MPT_REQUEST_AREA */
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
#define MPT_REQ_MEM_SIZE(mpt) (MPT_MAX_REQUESTS(mpt) * MPT_REQUEST_AREA)
Fix some of the previus changes 'better'. There's something strange going on with async events. They seem to be be treated differently for different Fusion implementations. Some will really tell you when it's okay to free the request that started them. Some won't. Very disconcerting. This is particularily bad when the chip (FC in this case) tells you in the reply that it's not a continuation reply, which means you can free the request that its associated with. However, if you do that, I've found that additional async event replies come back for that message context after you freed it. Very Bad Things Happen. Put in a reply register debounce. Warn about out of range context indices. Use more MPILIB defines where possible. Replace bzero with memset. Add tons more KASSERTS. Do a *lot* more request free list auditting and serial number usages. Get rid of the warning about the short IOC Facts Reply. Go back to 16 bits of context index. Do a lot more target state auditting as well. Make a tag out of not only the ioindex but the request index as well and worry less about keeping a full serial number.
2006-04-01 07:12:18 +00:00
#define MPT_CONTEXT_CB_SHIFT (16)
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
#define MPT_CBI(handle) (handle >> MPT_CONTEXT_CB_SHIFT)
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
#define MPT_CBI_TO_HID(cbi) ((cbi) << MPT_CONTEXT_CB_SHIFT)
#define MPT_CONTEXT_TO_CBI(x) \
(((x) >> MPT_CONTEXT_CB_SHIFT) & (MPT_NUM_REPLY_HANDLERS - 1))
Fix some of the previus changes 'better'. There's something strange going on with async events. They seem to be be treated differently for different Fusion implementations. Some will really tell you when it's okay to free the request that started them. Some won't. Very disconcerting. This is particularily bad when the chip (FC in this case) tells you in the reply that it's not a continuation reply, which means you can free the request that its associated with. However, if you do that, I've found that additional async event replies come back for that message context after you freed it. Very Bad Things Happen. Put in a reply register debounce. Warn about out of range context indices. Use more MPILIB defines where possible. Replace bzero with memset. Add tons more KASSERTS. Do a *lot* more request free list auditting and serial number usages. Get rid of the warning about the short IOC Facts Reply. Go back to 16 bits of context index. Do a lot more target state auditting as well. Make a tag out of not only the ioindex but the request index as well and worry less about keeping a full serial number.
2006-04-01 07:12:18 +00:00
#define MPT_CONTEXT_REQI_MASK 0xFFFF
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
#define MPT_CONTEXT_TO_REQI(x) ((x) & MPT_CONTEXT_REQI_MASK)
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
/*
* Convert a 32bit physical address returned from IOC to an
* offset into our reply frame memory or the kvm address needed
* to access the data. The returned address is only the low
* 32 bits, so mask our base physical address accordingly.
*/
#define MPT_REPLY_BADDR(x) \
(x << 1)
#define MPT_REPLY_OTOV(m, i) \
((void *)(&m->reply[i]))
#define MPT_DUMP_REPLY_FRAME(mpt, reply_frame) \
do { \
if (mpt->verbose >= MPT_PRT_DEBUG) \
mpt_dump_reply_frame(mpt, reply_frame); \
} while(0)
static __inline uint32_t mpt_pop_reply_queue(struct mpt_softc *mpt);
static __inline void mpt_free_reply(struct mpt_softc *mpt, uint32_t ptr);
/*
* Give the reply buffer back to the IOC after we have
* finished processing it.
*/
static __inline void
mpt_free_reply(struct mpt_softc *mpt, uint32_t ptr)
{
mpt_write(mpt, MPT_OFFSET_REPLY_Q, ptr);
}
/* Get a reply from the IOC */
static __inline uint32_t
mpt_pop_reply_queue(struct mpt_softc *mpt)
{
return mpt_read(mpt, MPT_OFFSET_REPLY_Q);
}
Fix some of the previus changes 'better'. There's something strange going on with async events. They seem to be be treated differently for different Fusion implementations. Some will really tell you when it's okay to free the request that started them. Some won't. Very disconcerting. This is particularily bad when the chip (FC in this case) tells you in the reply that it's not a continuation reply, which means you can free the request that its associated with. However, if you do that, I've found that additional async event replies come back for that message context after you freed it. Very Bad Things Happen. Put in a reply register debounce. Warn about out of range context indices. Use more MPILIB defines where possible. Replace bzero with memset. Add tons more KASSERTS. Do a *lot* more request free list auditting and serial number usages. Get rid of the warning about the short IOC Facts Reply. Go back to 16 bits of context index. Do a lot more target state auditting as well. Make a tag out of not only the ioindex but the request index as well and worry less about keeping a full serial number.
2006-04-01 07:12:18 +00:00
void
mpt_complete_request_chain(struct mpt_softc *, struct req_queue *, u_int);
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
/************************** Scatter Gather Managment **************************/
Do initial cut of SAS HBA support. These controllers (106X) seem to support automatically both SATA and SAS drives. The async SAS event handling we catch but ignore at present (so automagic attach/detach isn't hooked up yet). Do 64 bit PCI support- we can now work on systems with > 4GB of memory. Do large transfer support- we now can support up to reported chain depth, or the length of our request area. We simply allocate additional request elements when we would run out of room for chain lists. Tested on Ultra320, FC and SAS controllers on AMD64 and i386 platforms. There were no RAID cards available for me to regression test. The error recovery for this driver still is pretty bad.
2006-02-11 01:35:29 +00:00
/* MPT_RQSL- size of request frame, in bytes */
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
#define MPT_RQSL(mpt) (mpt->request_frame_size << 2)
Do initial cut of SAS HBA support. These controllers (106X) seem to support automatically both SATA and SAS drives. The async SAS event handling we catch but ignore at present (so automagic attach/detach isn't hooked up yet). Do 64 bit PCI support- we can now work on systems with > 4GB of memory. Do large transfer support- we now can support up to reported chain depth, or the length of our request area. We simply allocate additional request elements when we would run out of room for chain lists. Tested on Ultra320, FC and SAS controllers on AMD64 and i386 platforms. There were no RAID cards available for me to regression test. The error recovery for this driver still is pretty bad.
2006-02-11 01:35:29 +00:00
/* MPT_NSGL- how many SG entries can fit in a request frame size */
#define MPT_NSGL(mpt) (MPT_RQSL(mpt) / sizeof (SGE_IO_UNION))
/* MPT_NRFM- how many request frames can fit in each request alloc we make */
#define MPT_NRFM(mpt) (MPT_REQUEST_AREA / MPT_RQSL(mpt))
/*
* MPT_NSGL_FIRST- # of SG elements that can fit after
* an I/O request but still within the request frame.
* Do this safely based upon SGE_IO_UNION.
*
* Note that the first element is *within* the SCSI request.
*/
#define MPT_NSGL_FIRST(mpt) \
((MPT_RQSL(mpt) - sizeof (MSG_SCSI_IO_REQUEST) + sizeof (SGE_IO_UNION)) / \
sizeof (SGE_IO_UNION))
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
/***************************** IOC Initialization *****************************/
int mpt_reset(struct mpt_softc *, int /*reinit*/);
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
/****************************** Debugging ************************************/
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
typedef struct mpt_decode_entry {
char *name;
u_int value;
u_int mask;
} mpt_decode_entry_t;
int mpt_decode_value(mpt_decode_entry_t *table, u_int num_entries,
const char *name, u_int value, u_int *cur_column,
u_int wrap_point);
enum {
MPT_PRT_ALWAYS,
MPT_PRT_FATAL,
MPT_PRT_ERROR,
MPT_PRT_WARN,
MPT_PRT_INFO,
MPT_PRT_DEBUG,
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
MPT_PRT_DEBUG1,
MPT_PRT_DEBUG2,
MPT_PRT_DEBUG3,
Do initial cut of SAS HBA support. These controllers (106X) seem to support automatically both SATA and SAS drives. The async SAS event handling we catch but ignore at present (so automagic attach/detach isn't hooked up yet). Do 64 bit PCI support- we can now work on systems with > 4GB of memory. Do large transfer support- we now can support up to reported chain depth, or the length of our request area. We simply allocate additional request elements when we would run out of room for chain lists. Tested on Ultra320, FC and SAS controllers on AMD64 and i386 platforms. There were no RAID cards available for me to regression test. The error recovery for this driver still is pretty bad.
2006-02-11 01:35:29 +00:00
MPT_PRT_TRACE,
MPT_PRT_NONE=100
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
};
A large set of changes: + Add boatloads of KASSERTs and *really* check out more locking issues (to catch recursions when we actually go to real locking in CAM soon). The KASSERTs also caught lots of other issues like using commands that were put back on free lists, etc. + Target mode: role setting is derived directly from port capabilities. There is no need to set a role any more. Some target mode resources are allocated early on (ELS), but target command buffer allocation is deferred until the first lun enable. + Fix some breakages I introduced with target mode in that some commands are *repeating* commands. That is, the reply shows up but the command isn't really done (we don't free it). We still need to take it off the pending list because when we resubmit it, bad things then happen. + Fix more of the way that timed out commands and bus reset is done. The actual TMF response code was being ignored. + For SPI, honor BIOS settings. This doesn't quite fix the problems we've seen where we can't seem to (re)negotiate U320 on all drives but avoids it instead by letting us honor the BIOS settings. I'm sure this is not quite right and will have to change again soon.
2006-04-11 16:47:30 +00:00
#ifdef INVARIANTS
#define MPT_PRT_INVARIANT MPT_PRT_ALWAYS
#else
#define MPT_PRT_INVARIANT MPT_PRT_DEBUG
#endif
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
#if __FreeBSD_version > 500000
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
#define mpt_lprt(mpt, level, ...) \
do { \
if (level <= (mpt)->verbose) \
mpt_prt(mpt, __VA_ARGS__); \
} while (0)
#define mpt_lprtc(mpt, level, ...) \
do { \
if (level <= (mpt)->debug_level) \
mpt_prtc(mpt, __VA_ARGS__); \
} while (0)
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
#else
void mpt_lprt(struct mpt_softc *, int, const char *, ...)
__printflike(3, 4);
void mpt_lprtc(struct mpt_softc *, int, const char *, ...)
__printflike(3, 4);
#endif
void mpt_prt(struct mpt_softc *, const char *, ...)
__printflike(2, 3);
void mpt_prtc(struct mpt_softc *, const char *, ...)
__printflike(2, 3);
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
/**************************** Target Mode Related ***************************/
static __inline int mpt_cdblen(uint8_t, int);
static __inline int
mpt_cdblen(uint8_t cdb0, int maxlen)
{
int group = cdb0 >> 5;
switch (group) {
case 0:
return (6);
case 1:
return (10);
case 4:
case 5:
return (12);
default:
return (16);
}
}
#ifdef INVARIANTS
static __inline request_t * mpt_tag_2_req(struct mpt_softc *, uint32_t);
static __inline request_t *
mpt_tag_2_req(struct mpt_softc *mpt, uint32_t tag)
{
Fix some of the previus changes 'better'. There's something strange going on with async events. They seem to be be treated differently for different Fusion implementations. Some will really tell you when it's okay to free the request that started them. Some won't. Very disconcerting. This is particularily bad when the chip (FC in this case) tells you in the reply that it's not a continuation reply, which means you can free the request that its associated with. However, if you do that, I've found that additional async event replies come back for that message context after you freed it. Very Bad Things Happen. Put in a reply register debounce. Warn about out of range context indices. Use more MPILIB defines where possible. Replace bzero with memset. Add tons more KASSERTS. Do a *lot* more request free list auditting and serial number usages. Get rid of the warning about the short IOC Facts Reply. Go back to 16 bits of context index. Do a lot more target state auditting as well. Make a tag out of not only the ioindex but the request index as well and worry less about keeping a full serial number.
2006-04-01 07:12:18 +00:00
uint16_t rtg = (tag >> 18);
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
KASSERT(rtg < mpt->tgt_cmds_allocated, ("bad tag %d\n", tag));
KASSERT(mpt->tgt_cmd_ptrs, ("no cmd backpointer array"));
KASSERT(mpt->tgt_cmd_ptrs[rtg], ("no cmd backpointer"));
return (mpt->tgt_cmd_ptrs[rtg]);
}
A large set of changes: + Add boatloads of KASSERTs and *really* check out more locking issues (to catch recursions when we actually go to real locking in CAM soon). The KASSERTs also caught lots of other issues like using commands that were put back on free lists, etc. + Target mode: role setting is derived directly from port capabilities. There is no need to set a role any more. Some target mode resources are allocated early on (ELS), but target command buffer allocation is deferred until the first lun enable. + Fix some breakages I introduced with target mode in that some commands are *repeating* commands. That is, the reply shows up but the command isn't really done (we don't free it). We still need to take it off the pending list because when we resubmit it, bad things then happen. + Fix more of the way that timed out commands and bus reset is done. The actual TMF response code was being ignored. + For SPI, honor BIOS settings. This doesn't quite fix the problems we've seen where we can't seem to (re)negotiate U320 on all drives but avoids it instead by letting us honor the BIOS settings. I'm sure this is not quite right and will have to change again soon.
2006-04-11 16:47:30 +00:00
static __inline int
mpt_req_on_free_list(struct mpt_softc *, request_t *);
static __inline int
mpt_req_on_pending_list(struct mpt_softc *, request_t *);
static __inline void
mpt_req_spcl(struct mpt_softc *, request_t *, const char *, int);
static __inline void
mpt_req_not_spcl(struct mpt_softc *, request_t *, const char *, int);
/*
* Is request on freelist?
*/
static __inline int
mpt_req_on_free_list(struct mpt_softc *mpt, request_t *req)
{
request_t *lrq;
TAILQ_FOREACH(lrq, &mpt->request_free_list, links) {
if (lrq == req) {
return (1);
}
}
return (0);
}
/*
* Is request on pending list?
*/
static __inline int
mpt_req_on_pending_list(struct mpt_softc *mpt, request_t *req)
{
request_t *lrq;
TAILQ_FOREACH(lrq, &mpt->request_pending_list, links) {
if (lrq == req) {
return (1);
}
}
return (0);
}
/*
* Make sure that req *is* part of one of the special lists
*/
static __inline void
mpt_req_spcl(struct mpt_softc *mpt, request_t *req, const char *s, int line)
{
int i;
for (i = 0; i < mpt->els_cmds_allocated; i++) {
if (req == mpt->els_cmd_ptrs[i]) {
return;
}
}
for (i = 0; i < mpt->tgt_cmds_allocated; i++) {
if (req == mpt->tgt_cmd_ptrs[i]) {
return;
}
}
panic("%s(%d): req %p:%u function %x not in els or tgt ptrs\n",
s, line, req, req->serno,
((PTR_MSG_REQUEST_HEADER)req->req_vbuf)->Function);
}
/*
* Make sure that req is *not* part of one of the special lists.
*/
static __inline void
mpt_req_not_spcl(struct mpt_softc *mpt, request_t *req, const char *s, int line)
{
int i;
for (i = 0; i < mpt->els_cmds_allocated; i++) {
KASSERT(req != mpt->els_cmd_ptrs[i],
("%s(%d): req %p:%u func %x in els ptrs at ioindex %d\n",
s, line, req, req->serno,
((PTR_MSG_REQUEST_HEADER)req->req_vbuf)->Function, i));
}
for (i = 0; i < mpt->tgt_cmds_allocated; i++) {
KASSERT(req != mpt->tgt_cmd_ptrs[i],
("%s(%d): req %p:%u func %x in tgt ptrs at ioindex %d\n",
s, line, req, req->serno,
((PTR_MSG_REQUEST_HEADER)req->req_vbuf)->Function, i));
}
}
Some fairly major changes to this driver. A) Fibre Channel Target Mode support mostly works (SAS/SPI won't be too far behind). I'd say that this probably works just about as well as isp(4) does right now. Still, it and isp(4) and the whole target mode stack need a bit of tightening. B) The startup sequence has been changed so that after all attaches are done, a set of enable functions are called. The idea here is that the attaches do whatever needs to be done *prior* to a port being enabled and the enables do what need to be done for enabling stuff for a port after it's been enabled. This means that we also have events handled by their proper handlers as we start up. C) Conditional code that means that this driver goes back all the way to RELENG_4 in terms of support. D) Quite a lot of little nitty bug fixes- some discovered by doing RELENG_4 support. We've been living under Giant *waaaayyyyy* too long and it's made some of us (me) sloppy. E) Some shutdown hook stuff that makes sure we don't blow up during a reboot (like by the arrival of a new command from an initiator). There's been some testing and LINT checking, but not as complete as would be liked. Regression testing with Fusion RAID instances has not been possible. Caveat Emptor. Sponsored by: LSI-Logic.
2006-03-25 07:08:27 +00:00
#endif
typedef enum {
MPT_ABORT_TASK_SET=1234,
MPT_CLEAR_TASK_SET,
MPT_TARGET_RESET,
MPT_CLEAR_ACA,
MPT_TERMINATE_TASK,
MPT_NIL_TMT_VALUE=5678
} mpt_task_mgmt_t;
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
/**************************** Unclassified Routines ***************************/
void mpt_send_cmd(struct mpt_softc *mpt, request_t *req);
int mpt_recv_handshake_reply(struct mpt_softc *mpt,
size_t reply_len, void *reply);
int mpt_wait_req(struct mpt_softc *mpt, request_t *req,
mpt_req_state_t state, mpt_req_state_t mask,
int sleep_ok, int time_ms);
void mpt_enable_ints(struct mpt_softc *mpt);
void mpt_disable_ints(struct mpt_softc *mpt);
int mpt_attach(struct mpt_softc *mpt);
int mpt_shutdown(struct mpt_softc *mpt);
int mpt_detach(struct mpt_softc *mpt);
int mpt_send_handshake_cmd(struct mpt_softc *mpt,
size_t len, void *cmd);
request_t * mpt_get_request(struct mpt_softc *mpt, int sleep_ok);
void mpt_free_request(struct mpt_softc *mpt, request_t *req);
void mpt_intr(void *arg);
void mpt_check_doorbell(struct mpt_softc *mpt);
void mpt_dump_reply_frame(struct mpt_softc *mpt,
MSG_DEFAULT_REPLY *reply_frame);
void mpt_set_config_regs(struct mpt_softc *);
int mpt_issue_cfg_req(struct mpt_softc */*mpt*/, request_t */*req*/,
u_int /*Action*/, u_int /*PageVersion*/,
u_int /*PageLength*/, u_int /*PageNumber*/,
u_int /*PageType*/, uint32_t /*PageAddress*/,
bus_addr_t /*addr*/, bus_size_t/*len*/,
int /*sleep_ok*/, int /*timeout_ms*/);
int mpt_read_cfg_header(struct mpt_softc *, int /*PageType*/,
int /*PageNumber*/,
uint32_t /*PageAddress*/,
CONFIG_PAGE_HEADER *,
int /*sleep_ok*/, int /*timeout_ms*/);
int mpt_read_cfg_page(struct mpt_softc *t, int /*Action*/,
uint32_t /*PageAddress*/,
CONFIG_PAGE_HEADER *, size_t /*len*/,
int /*sleep_ok*/, int /*timeout_ms*/);
int mpt_write_cfg_page(struct mpt_softc *, int /*Action*/,
uint32_t /*PageAddress*/,
CONFIG_PAGE_HEADER *, size_t /*len*/,
int /*sleep_ok*/, int /*timeout_ms*/);
static __inline int
mpt_read_cur_cfg_page(struct mpt_softc *mpt, uint32_t PageAddress,
CONFIG_PAGE_HEADER *hdr, size_t len,
int sleep_ok, int timeout_ms)
{
return (mpt_read_cfg_page(mpt, MPI_CONFIG_ACTION_PAGE_READ_CURRENT,
PageAddress, hdr, len, sleep_ok, timeout_ms));
}
static __inline int
mpt_write_cur_cfg_page(struct mpt_softc *mpt, uint32_t PageAddress,
CONFIG_PAGE_HEADER *hdr, size_t len, int sleep_ok,
int timeout_ms)
{
return (mpt_write_cfg_page(mpt, MPI_CONFIG_ACTION_PAGE_WRITE_CURRENT,
PageAddress, hdr, len, sleep_ok,
timeout_ms));
}
Fix things so that: a) we don't believe what the board tells us all the time (if the BIOS hasn't run, port page 2 and port page 0 tend to be garbage) b) add the missing code to set parameters for the SPI cards. MFC after: 0 days
2002-09-01 23:07:09 +00:00
Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
/* mpt_debug.c functions */
void mpt_print_reply(void *vmsg);
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
void mpt_print_db(uint32_t mb);
Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
void mpt_print_config_reply(void *vmsg);
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
char *mpt_ioc_diag(uint32_t diag);
void mpt_req_state(mpt_req_state_t state);
Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
void mpt_print_config_request(void *vmsg);
void mpt_print_request(void *vmsg);
Massive overhaul of MPT Fusion driver: o Add timeout error recovery (from a thread context to avoid the deferral of other critical interrupts). o Properly recover commands across controller reset events. o Update the driver to handle events and status codes that have been added to the MPI spec since the driver was originally written. o Make the driver more modular to improve maintainability and support dynamic "personality" registration (e.g. SCSI Initiator, RAID, SAS, FC, etc). o Shorten and simplify the common I/O path to improve driver performance. o Add RAID volume and RAID member state/settings reporting. o Add periodic volume resynchronization status reporting. o Add support for sysctl tunable resync rate, member write cache enable, and volume transaction queue depth. Sponsored by ---------------- Avid Technologies Inc: SCSI error recovery, driver re-organization, update of MPI library headers, portions of dynamic personality registration, and misc bug fixes. Wheel Open Technologies: RAID event notification, RAID member pass-thru support, firmware upload/download support, enhanced RAID resync speed, portions of dynamic personality registration, and misc bug fixes. Detailed Changes ================ mpt.c mpt_cam.c mpt_raid.c mpt_pci.c: o Add support for personality modules. Each module exports load, and unload module scope methods as well as probe, attach, event, reset, shutdown, and detach per-device instance methods mpt.c mpt.h mpt_pci.c: o The driver now associates a callback function (via an index) with every transaction submitted to the controller. This allows the main interrupt handler to absolve itself of any knowledge of individual transaction/response types by simply calling the callback function "registered" for the transaction. We use a callback index instead of a callback function pointer in each requests so we can properly handle responses (e.g. event notifications) that are not associated with a transaction. Personality modules dynamically register their callbacks with the driver core to receive the callback index to use for their handlers. o Move the interrupt handler into mpt.c. The ISR algorithm is bus transport and OS independent and thus had no reason to be in mpt_pci.c. o Simplify configuration message reply handling by copying reply frame data for the requester and storing completion status in the original request structure. o Add the mpt_complete_request_chain() helper method and use it to implement reset handlers that must abort transactions. o Keep track of all pending requests on the new requests_pending_list in the softc. o Add default handlers to mpt.c to handle generic event notifications and controller reset activities. The event handler code is largely the same as in the original driver. The reset handler is new and terminates any pending transactions with a status code indicating the controller needs to be re-initialized. o Add some endian support to the driver. A complete audit is still required for this driver to have any hope of operating in a big-endian environment. o Use inttypes.h and __inline. Come closer to being style(9) compliant. o Remove extraneous use of typedefs. o Convert request state from a strict enumeration to a series of flags. This allows us to, for example, tag transactions that have timed-out while retaining the state that the transaction is still in-flight on the controller. o Add mpt_wait_req() which allows a caller to poll or sleep for the completion of a request. Use this to simplify and factor code out from many initialization routines. We also use this to sleep for task management request completions in our CAM timeout handler. mpt.c: o Correct a bug in the event handler where request structures were freed even if the request reply was marked as a continuation reply. Continuation replies indicate that the controller still owns the request and freeing these replies prematurely corrupted controller state. o Implement firmware upload and download. On controllers that do not have dedicated NVRAM (as in the Sun v20/v40z), the firmware image is downloaded to the controller by the system BIOS. This image occupies precious controller RAM space until the host driver fetches the image, reducing the number of concurrent I/Os the controller can processes. The uploaded image is used to re-program the controller during hard reset events since the controller cannot fetch the firmware on its own. Implementing this feature allows much higher queue depths when RAID volumes are configured. o Changed configuration page accessors to allow threads to sleep rather than busy wait for completion. o Removed hard coded data transfer sizes from configuration page routines so that RAID configuration page processing is possible. mpt_reg.h: o Move controller register definitions into a separate file. mpt.h: o Re-arrange includes to allow inlined functions to be defined in mpt.h. o Add reply, event, and reset handler definitions. o Add softc fields for handling timeout and controller reset recovery. mpt_cam.c: o Move mpt_freebsd.c to mpt_cam.c. Move all core functionality, such as event handling, into mpt.c leaving only CAM SCSI support here. o Revamp completion handler to provide correct CAM status for all currently defined SCSI MPI message result codes. o Register event and reset handlers with the MPT core. Modify the event handler to notify CAM of bus reset events. The controller reset handler will abort any transactions that have timed out. All other pending CAM transactions are correctly aborted by the core driver's reset handler. o Allocate a single request up front to perform task management operations. This guarantees that we can always perform a TMF operation even when the controller is saturated with other operations. The single request also serves as a perfect mechanism of guaranteeing that only a single TMF is in flight at a time - something that is required according to the MPT Fusion documentation. o Add a helper function for issuing task management requests to the controller. This is used to abort individual requests or perform a bus reset. o Modify the CAM XPT_BUS_RESET ccb handler to wait for and properly handle the status of the bus reset task management frame used to reset the bus. The previous code assumed that the reset request would always succeed. o Add timeout recovery support. When a timeout occurs, the timed-out request is added to a queue to be processed by our recovery thread and the thread is woken up. The recovery thread processes timed-out command serially, attempting first to abort them and then falling back to a bus reset if an abort fails. o Add calls to mpt_reset() to reset the controller if any handshake command, bus reset attempt or abort attempt fails due to a timeout. o Export a secondary "bus" to CAM that exposes all volume drive members as pass-thru devices, allowing CAM to perform proper speed negotiation to hidden devices. o Add a CAM async event handler tracking the AC_FOUND_DEVICE event. Use this to trigger calls to set the per-volume queue depth once the volume is fully registered with CAM. This is required to avoid hitting firmware limits on volume queue depth. Exceeding the limit causes the firmware to hang. mpt_cam.h: o Add several helper functions for interfacing to CAM and performing timeout recovery. mpt_pci.c: o Disable interrupts on the controller before registering and enabling interrupt delivery to the OS. Otherwise we risk receiving interrupts before the driver is ready to receive them. o Make use of compatibility macros that allow the driver to be compiled under 4.x and 5.x. mpt_raid.c: o Add a per-controller instance RAID thread to perform settings changes and query status (minimizes CPU busy wait loops). o Use a shutdown handler to disable "Member Write Cache Enable" (MWCE) setting for RAID arrays set to enable MWCE During Rebuild. o Change reply handler function signature to allow handlers to defer the deletion of reply frames. Use this to allow the event reply handler to queue up events that need to be acked if no resources are available to immediately ack an event. Queued events are processed in mpt_free_request() where resources are freed. This avoids a panic on resource shortage. o Parse and print out RAID controller capabilities during driver probe. o Define, allocate, and maintain RAID data structures for volumes, hidden member physical disks and spare disks. o Add dynamic sysctls for per-instance setting of the log level, array resync rate, array member cache enable, and volume queue depth. mpt_debug.c: o Add mpt_lprt and mpt_lprtc for printing diagnostics conditioned on a particular log level to aid in tracking down driver issues. o Add mpt_decode_value() which parses the bits in an integer value based on a parsing table (mask, value, name string, tuples). mpilib/*: o Update mpi library header files to latest distribution from LSI. Submitted by: gibbs Approved by: re
2005-07-10 15:05:39 +00:00
void mpt_print_scsi_io_request(MSG_SCSI_IO_REQUEST *msg);
Do initial cut of SAS HBA support. These controllers (106X) seem to support automatically both SATA and SAS drives. The async SAS event handling we catch but ignore at present (so automagic attach/detach isn't hooked up yet). Do 64 bit PCI support- we can now work on systems with > 4GB of memory. Do large transfer support- we now can support up to reported chain depth, or the length of our request area. We simply allocate additional request elements when we would run out of room for chain lists. Tested on Ultra320, FC and SAS controllers on AMD64 and i386 platforms. There were no RAID cards available for me to regression test. The error recovery for this driver still is pretty bad.
2006-02-11 01:35:29 +00:00
void mpt_dump_sgl(SGE_IO_UNION *se, int offset);
Add support for the LSI-Logic Fusion/MP architecture. This is an architecture that present a thing message passing interface to the OS. You can query as to how many ports and what kind are attached and enable them and so on. A less grand view is that this is just another way to package SCSI (SPI or FC) and FC-IP into a one-driver interface set. This driver support the following hardware: LSI FC909: Single channel, 1Gbps, Fibre Channel (FC-SCSI only) LSI FC929: Dual Channel, 1-2Gbps, Fibre Channel (FC-SCSI only) LSI 53c1020: Single Channel, Ultra4 (320M) (Untested) LSI 53c1030: Dual Channel, Ultra4 (320M) Currently it's in fair shape, but expect a lot of changes over the next few weeks as it stabilizes. Credits: The driver is mostly from some folks from Jeff Roberson's company- I've been slowly migrating it to broader support that I it came to me as. The hardware used in developing support came from: FC909: LSI-Logic, Advansys (now Connetix) FC929: LSI-Logic 53c1030: Antares Microsystems (they make a very fine board!) MFC after: 3 weeks
2002-08-11 23:34:20 +00:00
#endif /* _MPT_H_ */
Reference in New Issue Copy Permalink