freebsd-dev/sys/dev/mpt/mpt_pci.c

974 lines
26 KiB
C
Raw Normal View History

/*-
2002-08-30 02:35:31 +00:00
* PCI specific probe and attach routines for LSI Fusion Adapters
* FreeBSD Version.
*
* Copyright (c) 2000, 2001 by Greg Ansley
* Partially derived from Matt Jacob's ISP driver.
* Copyright (c) 1997, 1998, 1999, 2000, 2001, 2002 by Matthew Jacob
* Feral Software
* 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 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.
*/
/*-
* 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.
*
* Support from Chris Ellsworth in order to make SAS adapters work
* is gratefully acknowledged.
*
* Support from LSI-Logic has also gone a great deal toward making this a
* workable subsystem and is gratefully acknowledged.
*/
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) 2005, WHEEL Sp. z o.o.
* Copyright (c) 2004, 2005 Justin T. Gibbs
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are
* met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 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.
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.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
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 <dev/mpt/mpt.h>
#include <dev/mpt/mpt_cam.h>
#include <dev/mpt/mpt_raid.h>
#if __FreeBSD_version < 700000
#define pci_msix_count(x) 0
#define pci_msi_count(x) 0
#define pci_alloc_msi(x, y) 1
#define pci_alloc_msix(x, y) 1
#define pci_release_msi(x) do { ; } while (0)
#endif
#ifndef PCI_VENDOR_LSI
#define PCI_VENDOR_LSI 0x1000
#endif
#ifndef PCI_PRODUCT_LSI_FC909
#define PCI_PRODUCT_LSI_FC909 0x0620
#endif
2002-08-30 02:35:31 +00:00
#ifndef PCI_PRODUCT_LSI_FC909A
#define PCI_PRODUCT_LSI_FC909A 0x0621
#endif
#ifndef PCI_PRODUCT_LSI_FC919
#define PCI_PRODUCT_LSI_FC919 0x0624
#endif
#ifndef PCI_PRODUCT_LSI_FC929
#define PCI_PRODUCT_LSI_FC929 0x0622
#endif
#ifndef PCI_PRODUCT_LSI_FC929X
#define PCI_PRODUCT_LSI_FC929X 0x0626
#endif
#ifndef PCI_PRODUCT_LSI_FC919X
#define PCI_PRODUCT_LSI_FC919X 0x0628
#endif
#ifndef PCI_PRODUCT_LSI_FC7X04X
#define PCI_PRODUCT_LSI_FC7X04X 0x0640
#endif
2006-09-08 05:27:04 +00:00
#ifndef PCI_PRODUCT_LSI_FC646
#define PCI_PRODUCT_LSI_FC646 0x0646
#endif
#ifndef PCI_PRODUCT_LSI_1030
#define PCI_PRODUCT_LSI_1030 0x0030
#endif
#ifndef PCI_PRODUCT_LSI_SAS1064
#define PCI_PRODUCT_LSI_SAS1064 0x0050
#endif
#ifndef PCI_PRODUCT_LSI_SAS1064A
#define PCI_PRODUCT_LSI_SAS1064A 0x005C
#endif
#ifndef PCI_PRODUCT_LSI_SAS1064E
#define PCI_PRODUCT_LSI_SAS1064E 0x0056
#endif
#ifndef PCI_PRODUCT_LSI_SAS1066
#define PCI_PRODUCT_LSI_SAS1066 0x005E
#endif
#ifndef PCI_PRODUCT_LSI_SAS1066E
#define PCI_PRODUCT_LSI_SAS1066E 0x005A
#endif
#ifndef PCI_PRODUCT_LSI_SAS1068
#define PCI_PRODUCT_LSI_SAS1068 0x0054
#endif
#ifndef PCI_PRODUCT_LSI_SAS1068E
#define PCI_PRODUCT_LSI_SAS1068E 0x0058
#endif
#ifndef PCI_PRODUCT_LSI_SAS1078
#define PCI_PRODUCT_LSI_SAS1078 0x0062
#endif
#ifndef PCI_PRODUCT_LSI_SAS1078DE
#define PCI_PRODUCT_LSI_SAS1078DE 0x007C
#endif
#ifndef PCIM_CMD_SERRESPEN
#define PCIM_CMD_SERRESPEN 0x0100
#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
#define MPT_IO_BAR 0
#define MPT_MEM_BAR 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
static int mpt_pci_probe(device_t);
static int mpt_pci_attach(device_t);
static void mpt_free_bus_resources(struct mpt_softc *mpt);
static int mpt_pci_detach(device_t);
static int mpt_pci_shutdown(device_t);
static int mpt_dma_mem_alloc(struct mpt_softc *mpt);
static void mpt_dma_mem_free(struct mpt_softc *mpt);
static void mpt_read_config_regs(struct mpt_softc *mpt);
static void mpt_pci_intr(void *);
static device_method_t mpt_methods[] = {
/* Device interface */
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
DEVMETHOD(device_probe, mpt_pci_probe),
DEVMETHOD(device_attach, mpt_pci_attach),
DEVMETHOD(device_detach, mpt_pci_detach),
DEVMETHOD(device_shutdown, mpt_pci_shutdown),
{ 0, 0 }
};
static driver_t mpt_driver = {
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_methods, sizeof(struct mpt_softc)
};
static devclass_t mpt_devclass;
DRIVER_MODULE(mpt, pci, mpt_driver, mpt_devclass, 0, 0);
MODULE_DEPEND(mpt, pci, 1, 1, 1);
MODULE_VERSION(mpt, 1);
static 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
mpt_pci_probe(device_t dev)
{
char *desc;
if (pci_get_vendor(dev) != PCI_VENDOR_LSI) {
return (ENXIO);
}
switch ((pci_get_device(dev) & ~1)) {
case PCI_PRODUCT_LSI_FC909:
desc = "LSILogic FC909 FC Adapter";
break;
2002-08-30 02:35:31 +00:00
case PCI_PRODUCT_LSI_FC909A:
desc = "LSILogic FC909A FC Adapter";
break;
case PCI_PRODUCT_LSI_FC919:
desc = "LSILogic FC919 FC Adapter";
break;
case PCI_PRODUCT_LSI_FC929:
2006-09-08 05:27:04 +00:00
desc = "Dual LSILogic FC929 FC Adapter";
break;
case PCI_PRODUCT_LSI_FC919X:
2006-09-08 05:27:04 +00:00
desc = "LSILogic FC919 FC PCI-X Adapter";
break;
case PCI_PRODUCT_LSI_FC929X:
2006-09-08 05:27:04 +00:00
desc = "Dual LSILogic FC929X 2Gb/s FC PCI-X Adapter";
break;
case PCI_PRODUCT_LSI_FC646:
desc = "Dual LSILogic FC7X04X 4Gb/s FC PCI-Express Adapter";
break;
case PCI_PRODUCT_LSI_FC7X04X:
2006-09-08 05:27:04 +00:00
desc = "Dual LSILogic FC7X04X 4Gb/s FC PCI-X Adapter";
break;
case PCI_PRODUCT_LSI_1030:
desc = "LSILogic 1030 Ultra4 Adapter";
break;
case PCI_PRODUCT_LSI_SAS1064:
case PCI_PRODUCT_LSI_SAS1064A:
case PCI_PRODUCT_LSI_SAS1064E:
case PCI_PRODUCT_LSI_SAS1066:
case PCI_PRODUCT_LSI_SAS1066E:
case PCI_PRODUCT_LSI_SAS1068:
case PCI_PRODUCT_LSI_SAS1068E:
case PCI_PRODUCT_LSI_SAS1078:
case PCI_PRODUCT_LSI_SAS1078DE:
2006-09-08 05:27:04 +00:00
desc = "LSILogic SAS/SATA Adapter";
break;
default:
return (ENXIO);
}
device_set_desc(dev, desc);
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
return (0);
}
#if __FreeBSD_version < 500000
static void
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_set_options(struct mpt_softc *mpt)
{
int bitmap;
bitmap = 0;
if (getenv_int("mpt_disable", &bitmap)) {
if (bitmap & (1 << mpt->unit)) {
mpt->disabled = 1;
}
}
bitmap = 0;
if (getenv_int("mpt_debug", &bitmap)) {
if (bitmap & (1 << mpt->unit)) {
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->verbose = MPT_PRT_DEBUG;
}
}
bitmap = 0;
if (getenv_int("mpt_debug1", &bitmap)) {
if (bitmap & (1 << mpt->unit)) {
mpt->verbose = MPT_PRT_DEBUG1;
}
}
bitmap = 0;
if (getenv_int("mpt_debug2", &bitmap)) {
if (bitmap & (1 << mpt->unit)) {
mpt->verbose = MPT_PRT_DEBUG2;
}
}
bitmap = 0;
if (getenv_int("mpt_debug3", &bitmap)) {
if (bitmap & (1 << mpt->unit)) {
mpt->verbose = MPT_PRT_DEBUG3;
}
}
mpt->cfg_role = MPT_ROLE_DEFAULT;
bitmap = 0;
if (getenv_int("mpt_nil_role", &bitmap)) {
if (bitmap & (1 << mpt->unit)) {
mpt->cfg_role = 0;
}
mpt->do_cfg_role = 1;
}
bitmap = 0;
if (getenv_int("mpt_tgt_role", &bitmap)) {
if (bitmap & (1 << mpt->unit)) {
mpt->cfg_role |= MPT_ROLE_TARGET;
}
mpt->do_cfg_role = 1;
}
bitmap = 0;
if (getenv_int("mpt_ini_role", &bitmap)) {
if (bitmap & (1 << mpt->unit)) {
mpt->cfg_role |= MPT_ROLE_INITIATOR;
}
mpt->do_cfg_role = 1;
}
mpt->msi_enable = 0;
}
#else
static void
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_set_options(struct mpt_softc *mpt)
{
int tval;
tval = 0;
if (resource_int_value(device_get_name(mpt->dev),
device_get_unit(mpt->dev), "disable", &tval) == 0 && tval != 0) {
mpt->disabled = 1;
}
tval = 0;
if (resource_int_value(device_get_name(mpt->dev),
device_get_unit(mpt->dev), "debug", &tval) == 0 && tval != 0) {
mpt->verbose = tval;
}
tval = -1;
if (resource_int_value(device_get_name(mpt->dev),
device_get_unit(mpt->dev), "role", &tval) == 0 && tval >= 0 &&
tval <= 3) {
mpt->cfg_role = tval;
mpt->do_cfg_role = 1;
}
tval = 0;
mpt->msi_enable = 0;
if (resource_int_value(device_get_name(mpt->dev),
device_get_unit(mpt->dev), "msi_enable", &tval) == 0 && tval == 1) {
mpt->msi_enable = 1;
}
}
#endif
static void
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_link_peer(struct mpt_softc *mpt)
{
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_softc *mpt2;
if (mpt->unit == 0) {
return;
}
/*
* XXX: depends on probe order
*/
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
mpt2 = (struct mpt_softc *)devclass_get_softc(mpt_devclass,mpt->unit-1);
if (mpt2 == NULL) {
return;
}
if (pci_get_vendor(mpt2->dev) != pci_get_vendor(mpt->dev)) {
return;
}
if (pci_get_device(mpt2->dev) != pci_get_device(mpt->dev)) {
return;
}
mpt->mpt2 = mpt2;
mpt2->mpt2 = mpt;
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 (mpt->verbose >= MPT_PRT_DEBUG) {
mpt_prt(mpt, "linking with peer (mpt%d)\n",
device_get_unit(mpt2->dev));
}
}
static void
mpt_unlink_peer(struct mpt_softc *mpt)
{
if (mpt->mpt2) {
mpt->mpt2->mpt2 = NULL;
}
}
static 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
mpt_pci_attach(device_t dev)
{
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_softc *mpt;
int iqd;
uint32_t data, cmd;
/* Allocate the softc structure */
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 = (struct mpt_softc*)device_get_softc(dev);
if (mpt == NULL) {
device_printf(dev, "cannot allocate softc\n");
return (ENOMEM);
}
memset(mpt, 0, sizeof(struct mpt_softc));
switch ((pci_get_device(dev) & ~1)) {
case PCI_PRODUCT_LSI_FC909:
2002-08-30 02:35:31 +00:00
case PCI_PRODUCT_LSI_FC909A:
case PCI_PRODUCT_LSI_FC919:
case PCI_PRODUCT_LSI_FC929:
case PCI_PRODUCT_LSI_FC919X:
2006-09-08 05:27:04 +00:00
case PCI_PRODUCT_LSI_FC646:
case PCI_PRODUCT_LSI_FC7X04X:
mpt->is_fc = 1;
break;
case PCI_PRODUCT_LSI_SAS1064:
case PCI_PRODUCT_LSI_SAS1064A:
case PCI_PRODUCT_LSI_SAS1064E:
case PCI_PRODUCT_LSI_SAS1066:
case PCI_PRODUCT_LSI_SAS1066E:
case PCI_PRODUCT_LSI_SAS1068:
case PCI_PRODUCT_LSI_SAS1068E:
case PCI_PRODUCT_LSI_SAS1078:
case PCI_PRODUCT_LSI_SAS1078DE:
mpt->is_sas = 1;
break;
default:
mpt->is_spi = 1;
break;
}
mpt->dev = dev;
mpt->unit = device_get_unit(dev);
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->raid_resync_rate = MPT_RAID_RESYNC_RATE_DEFAULT;
mpt->raid_mwce_setting = MPT_RAID_MWCE_DEFAULT;
mpt->raid_queue_depth = MPT_RAID_QUEUE_DEPTH_DEFAULT;
mpt->verbose = MPT_PRT_NONE;
mpt->role = MPT_ROLE_NONE;
mpt_set_options(mpt);
if (mpt->verbose == MPT_PRT_NONE) {
mpt->verbose = MPT_PRT_WARN;
/* Print INFO level (if any) if bootverbose is set */
mpt->verbose += (bootverbose != 0)? 1 : 0;
}
/* Make sure memory access decoders are enabled */
cmd = pci_read_config(dev, PCIR_COMMAND, 2);
if ((cmd & PCIM_CMD_MEMEN) == 0) {
device_printf(dev, "Memory accesses disabled");
return (ENXIO);
}
/*
* Make sure that SERR, PERR, WRITE INVALIDATE and BUSMASTER are set.
*/
cmd |=
PCIM_CMD_SERRESPEN | PCIM_CMD_PERRESPEN |
PCIM_CMD_BUSMASTEREN | PCIM_CMD_MWRICEN;
pci_write_config(dev, PCIR_COMMAND, cmd, 2);
/*
* Make sure we've disabled the ROM.
*/
data = pci_read_config(dev, PCIR_BIOS, 4);
data &= ~1;
pci_write_config(dev, PCIR_BIOS, data, 4);
/*
* Is this part a dual?
* If so, link with our partner (around yet)
*/
if ((pci_get_device(dev) & ~1) == PCI_PRODUCT_LSI_FC929 ||
2006-09-08 05:27:04 +00:00
(pci_get_device(dev) & ~1) == PCI_PRODUCT_LSI_FC646 ||
(pci_get_device(dev) & ~1) == PCI_PRODUCT_LSI_FC7X04X ||
(pci_get_device(dev) & ~1) == PCI_PRODUCT_LSI_1030) {
mpt_link_peer(mpt);
}
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
/*
* Set up register access. PIO mode is required for
* certain reset operations (but must be disabled for
* some cards otherwise).
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->pci_pio_rid = PCIR_BAR(MPT_IO_BAR);
mpt->pci_pio_reg = bus_alloc_resource(dev, SYS_RES_IOPORT,
&mpt->pci_pio_rid, 0, ~0, 0, RF_ACTIVE);
if (mpt->pci_pio_reg == NULL) {
device_printf(dev, "unable to map registers in PIO mode\n");
goto bad;
}
mpt->pci_pio_st = rman_get_bustag(mpt->pci_pio_reg);
mpt->pci_pio_sh = rman_get_bushandle(mpt->pci_pio_reg);
/* Allocate kernel virtual memory for the 9x9's Mem0 region */
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->pci_mem_rid = PCIR_BAR(MPT_MEM_BAR);
mpt->pci_reg = bus_alloc_resource(dev, SYS_RES_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
&mpt->pci_mem_rid, 0, ~0, 0, RF_ACTIVE);
if (mpt->pci_reg == NULL) {
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_printf(dev, "Unable to memory map registers.\n");
if (mpt->is_sas) {
device_printf(dev, "Giving Up.\n");
goto bad;
}
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_printf(dev, "Falling back to PIO mode.\n");
mpt->pci_st = mpt->pci_pio_st;
mpt->pci_sh = mpt->pci_pio_sh;
} else {
mpt->pci_st = rman_get_bustag(mpt->pci_reg);
mpt->pci_sh = rman_get_bushandle(mpt->pci_reg);
}
/* Get a handle to the interrupt */
iqd = 0;
if (mpt->msi_enable) {
/*
* First try to alloc an MSI-X message. If that
* fails, then try to alloc an MSI message instead.
*/
if (pci_msix_count(dev) == 1) {
mpt->pci_msi_count = 1;
if (pci_alloc_msix(dev, &mpt->pci_msi_count) == 0) {
iqd = 1;
} else {
mpt->pci_msi_count = 0;
}
}
if (iqd == 0 && pci_msi_count(dev) == 1) {
mpt->pci_msi_count = 1;
if (pci_alloc_msi(dev, &mpt->pci_msi_count) == 0) {
iqd = 1;
} else {
mpt->pci_msi_count = 0;
}
}
}
mpt->pci_irq = bus_alloc_resource_any(dev, SYS_RES_IRQ, &iqd,
RF_ACTIVE | RF_SHAREABLE);
if (mpt->pci_irq == NULL) {
device_printf(dev, "could not allocate interrupt\n");
goto bad;
}
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_LOCK_SETUP(mpt);
/* Disable interrupts at the part */
mpt_disable_ints(mpt);
/* Register the interrupt handler */
if (mpt_setup_intr(dev, mpt->pci_irq, MPT_IFLAGS, NULL, mpt_pci_intr,
mpt, &mpt->ih)) {
device_printf(dev, "could not setup interrupt\n");
goto bad;
}
/* Allocate dma 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
/* XXX JGibbs -Should really be done based on IOCFacts. */
if (mpt_dma_mem_alloc(mpt)) {
mpt_prt(mpt, "Could not allocate DMA memory\n");
goto bad;
}
/*
* Save the PCI config register values
*
* Hard resets are known to screw up the BAR for diagnostic
* memory accesses (Mem1).
*
* Using Mem1 is known to make the chip stop responding to
* configuration space transfers, so we need to save it now
*/
mpt_read_config_regs(mpt);
/*
* Disable PIO until we need it
*/
if (mpt->is_sas) {
pci_disable_io(dev, SYS_RES_IOPORT);
}
/* Initialize the hardware */
if (mpt->disabled == 0) {
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 (mpt_attach(mpt) != 0) {
goto bad;
}
} else {
mpt_prt(mpt, "device disabled at user request\n");
goto bad;
}
mpt->eh = EVENTHANDLER_REGISTER(shutdown_post_sync, mpt_pci_shutdown,
dev, SHUTDOWN_PRI_DEFAULT);
if (mpt->eh == NULL) {
mpt_prt(mpt, "shutdown event registration failed\n");
(void) mpt_detach(mpt);
goto bad;
}
return (0);
bad:
mpt_dma_mem_free(mpt);
mpt_free_bus_resources(mpt);
mpt_unlink_peer(mpt);
MPT_LOCK_DESTROY(mpt);
/*
* but return zero to preserve unit numbering
*/
return (0);
}
/*
* Free bus resources
*/
static void
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_free_bus_resources(struct mpt_softc *mpt)
{
if (mpt->ih) {
bus_teardown_intr(mpt->dev, mpt->pci_irq, mpt->ih);
mpt->ih = 0;
}
if (mpt->pci_irq) {
bus_release_resource(mpt->dev, SYS_RES_IRQ,
mpt->pci_msi_count ? 1 : 0, mpt->pci_irq);
mpt->pci_irq = 0;
}
if (mpt->pci_msi_count) {
pci_release_msi(mpt->dev);
mpt->pci_msi_count = 0;
}
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 (mpt->pci_pio_reg) {
bus_release_resource(mpt->dev, SYS_RES_IOPORT, mpt->pci_pio_rid,
mpt->pci_pio_reg);
mpt->pci_pio_reg = 0;
}
if (mpt->pci_reg) {
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
bus_release_resource(mpt->dev, SYS_RES_MEMORY, mpt->pci_mem_rid,
mpt->pci_reg);
mpt->pci_reg = 0;
}
MPT_LOCK_DESTROY(mpt);
}
/*
* Disconnect ourselves from the system.
*/
static 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
mpt_pci_detach(device_t dev)
{
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_softc *mpt;
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 = (struct mpt_softc*)device_get_softc(dev);
if (mpt) {
mpt_disable_ints(mpt);
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_detach(mpt);
mpt_reset(mpt, /*reinit*/FALSE);
mpt_dma_mem_free(mpt);
mpt_free_bus_resources(mpt);
mpt_raid_free_mem(mpt);
if (mpt->eh != NULL) {
EVENTHANDLER_DEREGISTER(shutdown_post_sync, mpt->eh);
}
}
return(0);
}
/*
* Disable the hardware
*/
static 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
mpt_pci_shutdown(device_t dev)
{
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_softc *mpt;
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 = (struct mpt_softc *)device_get_softc(dev);
if (mpt) {
int r;
r = mpt_shutdown(mpt);
return (r);
}
return(0);
}
static 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
mpt_dma_mem_alloc(struct mpt_softc *mpt)
{
int i, error, nsegs;
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
uint8_t *vptr;
uint32_t pptr, end;
size_t len;
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_map_info mi;
/* Check if we alreay have allocated the reply memory */
if (mpt->reply_phys != 0) {
return 0;
}
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
len = sizeof (request_t) * MPT_MAX_REQUESTS(mpt);
#ifdef RELENG_4
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->request_pool = (request_t *)malloc(len, M_DEVBUF, M_WAITOK);
if (mpt->request_pool == NULL) {
mpt_prt(mpt, "cannot allocate request pool\n");
return (1);
}
memset(mpt->request_pool, 0, len);
#else
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->request_pool = (request_t *)malloc(len, M_DEVBUF, M_WAITOK|M_ZERO);
if (mpt->request_pool == NULL) {
mpt_prt(mpt, "cannot allocate request pool\n");
return (1);
}
#endif
/*
* Create a parent dma tag for this device.
*
* Align at byte boundaries,
* Limit to 32-bit addressing for request/reply queues.
*/
if (mpt_dma_tag_create(mpt, /*parent*/bus_get_dma_tag(mpt->dev),
/*alignment*/1, /*boundary*/0, /*lowaddr*/BUS_SPACE_MAXADDR,
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
/*highaddr*/BUS_SPACE_MAXADDR, /*filter*/NULL, /*filterarg*/NULL,
/*maxsize*/BUS_SPACE_MAXSIZE_32BIT,
/*nsegments*/BUS_SPACE_MAXSIZE_32BIT,
/*maxsegsz*/BUS_SPACE_UNRESTRICTED, /*flags*/0,
&mpt->parent_dmat) != 0) {
mpt_prt(mpt, "cannot create parent dma tag\n");
return (1);
}
/* Create a child tag for reply buffers */
if (mpt_dma_tag_create(mpt, mpt->parent_dmat, PAGE_SIZE, 0,
BUS_SPACE_MAXADDR_32BIT, BUS_SPACE_MAXADDR,
NULL, NULL, 2 * PAGE_SIZE, 1, BUS_SPACE_MAXSIZE_32BIT, 0,
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->reply_dmat) != 0) {
mpt_prt(mpt, "cannot create a dma tag for replies\n");
return (1);
}
/* Allocate some DMA accessable memory for replies */
if (bus_dmamem_alloc(mpt->reply_dmat, (void **)&mpt->reply,
BUS_DMA_NOWAIT, &mpt->reply_dmap) != 0) {
mpt_prt(mpt, "cannot allocate %lu bytes of reply memory\n",
(u_long) (2 * PAGE_SIZE));
return (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
mi.mpt = mpt;
mi.error = 0;
/* Load and lock it into "bus space" */
bus_dmamap_load(mpt->reply_dmat, mpt->reply_dmap, mpt->reply,
2 * PAGE_SIZE, mpt_map_rquest, &mi, 0);
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 (mi.error) {
mpt_prt(mpt, "error %d loading dma map for DMA reply queue\n",
mi.error);
return (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
mpt->reply_phys = mi.phys;
/* Create a child tag for data buffers */
/*
* XXX: we should say that nsegs is 'unrestricted, but that
* XXX: tickles a horrible bug in the busdma code. Instead,
* XXX: we'll derive a reasonable segment limit from MAXPHYS
*/
nsegs = (MAXPHYS / PAGE_SIZE) + 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
if (mpt_dma_tag_create(mpt, mpt->parent_dmat, 1,
0, BUS_SPACE_MAXADDR, BUS_SPACE_MAXADDR,
NULL, NULL, MAXBSIZE, nsegs, BUS_SPACE_MAXSIZE_32BIT, 0,
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->buffer_dmat) != 0) {
mpt_prt(mpt, "cannot create a dma tag for data buffers\n");
return (1);
}
/* Create a child tag for request buffers */
if (mpt_dma_tag_create(mpt, mpt->parent_dmat, PAGE_SIZE, 0,
BUS_SPACE_MAXADDR_32BIT, BUS_SPACE_MAXADDR,
NULL, NULL, MPT_REQ_MEM_SIZE(mpt), 1, BUS_SPACE_MAXSIZE_32BIT, 0,
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->request_dmat) != 0) {
mpt_prt(mpt, "cannot create a dma tag for requests\n");
return (1);
}
/* Allocate some DMA accessable memory for requests */
if (bus_dmamem_alloc(mpt->request_dmat, (void **)&mpt->request,
BUS_DMA_NOWAIT, &mpt->request_dmap) != 0) {
mpt_prt(mpt, "cannot allocate %d bytes of request memory\n",
MPT_REQ_MEM_SIZE(mpt));
return (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
mi.mpt = mpt;
mi.error = 0;
/* Load and lock it into "bus space" */
bus_dmamap_load(mpt->request_dmat, mpt->request_dmap, mpt->request,
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_REQ_MEM_SIZE(mpt), mpt_map_rquest, &mi, 0);
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 (mi.error) {
mpt_prt(mpt, "error %d loading dma map for DMA request queue\n",
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
mi.error);
return (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
mpt->request_phys = mi.phys;
/*
* Now create per-request dma maps
*/
i = 0;
pptr = mpt->request_phys;
vptr = mpt->request;
end = pptr + MPT_REQ_MEM_SIZE(mpt);
while(pptr < end) {
request_t *req = &mpt->request_pool[i];
req->index = i++;
/* Store location of Request Data */
req->req_pbuf = pptr;
req->req_vbuf = vptr;
pptr += MPT_REQUEST_AREA;
vptr += MPT_REQUEST_AREA;
req->sense_pbuf = (pptr - MPT_SENSE_SIZE);
req->sense_vbuf = (vptr - MPT_SENSE_SIZE);
error = bus_dmamap_create(mpt->buffer_dmat, 0, &req->dmap);
if (error) {
mpt_prt(mpt, "error %d creating per-cmd DMA maps\n",
error);
return (1);
}
}
return (0);
}
/* Deallocate memory that was allocated by mpt_dma_mem_alloc
*/
static void
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_dma_mem_free(struct mpt_softc *mpt)
{
int i;
/* Make sure we aren't double destroying */
if (mpt->reply_dmat == 0) {
mpt_lprt(mpt, MPT_PRT_DEBUG, "already released dma memory\n");
return;
}
for (i = 0; i < MPT_MAX_REQUESTS(mpt); i++) {
bus_dmamap_destroy(mpt->buffer_dmat, mpt->request_pool[i].dmap);
}
bus_dmamap_unload(mpt->request_dmat, mpt->request_dmap);
bus_dmamem_free(mpt->request_dmat, mpt->request, mpt->request_dmap);
bus_dma_tag_destroy(mpt->request_dmat);
bus_dma_tag_destroy(mpt->buffer_dmat);
bus_dmamap_unload(mpt->reply_dmat, mpt->reply_dmap);
bus_dmamem_free(mpt->reply_dmat, mpt->reply, mpt->reply_dmap);
bus_dma_tag_destroy(mpt->reply_dmat);
bus_dma_tag_destroy(mpt->parent_dmat);
mpt->reply_dmat = 0;
free(mpt->request_pool, M_DEVBUF);
mpt->request_pool = 0;
}
/* Reads modifiable (via PCI transactions) config registers */
static void
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_read_config_regs(struct mpt_softc *mpt)
{
mpt->pci_cfg.Command = pci_read_config(mpt->dev, PCIR_COMMAND, 2);
mpt->pci_cfg.LatencyTimer_LineSize =
pci_read_config(mpt->dev, PCIR_CACHELNSZ, 2);
mpt->pci_cfg.IO_BAR = pci_read_config(mpt->dev, PCIR_BAR(0), 4);
mpt->pci_cfg.Mem0_BAR[0] = pci_read_config(mpt->dev, PCIR_BAR(1), 4);
mpt->pci_cfg.Mem0_BAR[1] = pci_read_config(mpt->dev, PCIR_BAR(2), 4);
mpt->pci_cfg.Mem1_BAR[0] = pci_read_config(mpt->dev, PCIR_BAR(3), 4);
mpt->pci_cfg.Mem1_BAR[1] = pci_read_config(mpt->dev, PCIR_BAR(4), 4);
mpt->pci_cfg.ROM_BAR = pci_read_config(mpt->dev, PCIR_BIOS, 4);
mpt->pci_cfg.IntLine = pci_read_config(mpt->dev, PCIR_INTLINE, 1);
mpt->pci_cfg.PMCSR = pci_read_config(mpt->dev, 0x44, 4);
}
/* Sets modifiable config registers */
void
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_set_config_regs(struct mpt_softc *mpt)
{
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
uint32_t val;
#define MPT_CHECK(reg, offset, size) \
val = pci_read_config(mpt->dev, offset, size); \
if (mpt->pci_cfg.reg != val) { \
mpt_prt(mpt, \
"Restoring " #reg " to 0x%X from 0x%X\n", \
mpt->pci_cfg.reg, val); \
}
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 (mpt->verbose >= MPT_PRT_DEBUG) {
MPT_CHECK(Command, PCIR_COMMAND, 2);
MPT_CHECK(LatencyTimer_LineSize, PCIR_CACHELNSZ, 2);
MPT_CHECK(IO_BAR, PCIR_BAR(0), 4);
MPT_CHECK(Mem0_BAR[0], PCIR_BAR(1), 4);
MPT_CHECK(Mem0_BAR[1], PCIR_BAR(2), 4);
MPT_CHECK(Mem1_BAR[0], PCIR_BAR(3), 4);
MPT_CHECK(Mem1_BAR[1], PCIR_BAR(4), 4);
MPT_CHECK(ROM_BAR, PCIR_BIOS, 4);
MPT_CHECK(IntLine, PCIR_INTLINE, 1);
MPT_CHECK(PMCSR, 0x44, 4);
}
#undef MPT_CHECK
pci_write_config(mpt->dev, PCIR_COMMAND, mpt->pci_cfg.Command, 2);
pci_write_config(mpt->dev, PCIR_CACHELNSZ,
mpt->pci_cfg.LatencyTimer_LineSize, 2);
pci_write_config(mpt->dev, PCIR_BAR(0), mpt->pci_cfg.IO_BAR, 4);
pci_write_config(mpt->dev, PCIR_BAR(1), mpt->pci_cfg.Mem0_BAR[0], 4);
pci_write_config(mpt->dev, PCIR_BAR(2), mpt->pci_cfg.Mem0_BAR[1], 4);
pci_write_config(mpt->dev, PCIR_BAR(3), mpt->pci_cfg.Mem1_BAR[0], 4);
pci_write_config(mpt->dev, PCIR_BAR(4), mpt->pci_cfg.Mem1_BAR[1], 4);
pci_write_config(mpt->dev, PCIR_BIOS, mpt->pci_cfg.ROM_BAR, 4);
pci_write_config(mpt->dev, PCIR_INTLINE, mpt->pci_cfg.IntLine, 1);
pci_write_config(mpt->dev, 0x44, mpt->pci_cfg.PMCSR, 4);
}
static void
mpt_pci_intr(void *arg)
{
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_softc *mpt;
mpt = (struct mpt_softc *)arg;
MPT_LOCK(mpt);
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_intr(mpt);
MPT_UNLOCK(mpt);
}