d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
0692579bf3
freebsd-skq/sys/dev/nvme/nvme_private.h

475 lines
14 KiB
C
Raw Normal View History

This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
/*-
* Copyright (C) 2012 Intel Corporation
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* $FreeBSD$
*/
#ifndef __NVME_PRIVATE_H__
#define __NVME_PRIVATE_H__
#include <sys/param.h>
#include <sys/kernel.h>
#include <sys/lock.h>
#include <sys/malloc.h>
#include <sys/mutex.h>
#include <sys/rman.h>
#include <sys/systm.h>
Add handling for controller fatal status (csts.cfs). On any I/O timeout, check for csts.cfs==1. If set, the controller is reporting fatal status and we reset the controller immediately, rather than trying to abort the timed out command. This changeset also includes deferring the controller start portion of the reset to a separate task. This ensures we are always performing a controller start operation from a consistent context. Sponsored by: Intel Reviewed by: carl
2013-03-26 19:58:17 +00:00
#include <sys/taskqueue.h>
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
Add struct nvme_request object which contains all of the parameters passed from an NVMe consumer. This allows us to mostly build NVMe command buffers without holding the qpair lock, and also allows for future queueing of nvme_request objects in cases where the submission queue is full and no nvme_tracker objects are available. Sponsored by: Intel
2012-10-18 00:38:28 +00:00
#include <vm/uma.h>
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
#include <machine/bus.h>
#include "nvme.h"
#define DEVICE2SOFTC(dev) ((struct nvme_controller *) device_get_softc(dev))
MALLOC_DECLARE(M_NVME);
#define CHATHAM2
#ifdef CHATHAM2
#define CHATHAM_PCI_ID 0x20118086
#define CHATHAM_CONTROL_BAR 0
#endif
Add PCI device ID for 8-channel IDT NVMe controller, and clarify that the previously defined IDT PCI device ID was for a 32-channel controller. Submitted by: Joe Golio <joseph.golio@isilon.com>
2012-12-06 15:36:24 +00:00
#define IDT32_PCI_ID 0x80d0111d /* 32 channel board */
#define IDT8_PCI_ID 0x80d2111d /* 8 channel board */
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
Merge struct nvme_prp_list into struct nvme_tracker. This simplifies the driver significantly where it is constructing commands to be submitted to hardware. By reducing the number of PRPs (NVMe parlance for SGE) from 128 to 32, it ensures we do not allocate too much memory for more common smaller I/O sizes, while still supporting up to 128KB I/O sizes. This also paves the way for pre-allocation of nvme_tracker objects for each queue which will simplify the I/O path even further. Sponsored by: Intel
2012-10-18 00:37:11 +00:00
#define NVME_MAX_PRP_LIST_ENTRIES (32)
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
/*
* For commands requiring more than 2 PRP entries, one PRP will be
* embedded in the command (prp1), and the rest of the PRP entries
* will be in a list pointed to by the command (prp2). This means
Merge struct nvme_prp_list into struct nvme_tracker. This simplifies the driver significantly where it is constructing commands to be submitted to hardware. By reducing the number of PRPs (NVMe parlance for SGE) from 128 to 32, it ensures we do not allocate too much memory for more common smaller I/O sizes, while still supporting up to 128KB I/O sizes. This also paves the way for pre-allocation of nvme_tracker objects for each queue which will simplify the I/O path even further. Sponsored by: Intel
2012-10-18 00:37:11 +00:00
* that real max number of PRP entries we support is 32+1, which
* results in a max xfer size of 32*PAGE_SIZE.
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
*/
#define NVME_MAX_XFER_SIZE NVME_MAX_PRP_LIST_ENTRIES * PAGE_SIZE
Preallocate a limited number of nvme_tracker objects per qpair, rather than dynamically creating them at runtime. Sponsored by: Intel
2012-10-18 00:44:39 +00:00
#define NVME_ADMIN_TRACKERS (16)
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
#define NVME_ADMIN_ENTRIES (128)
/* min and max are defined in admin queue attributes section of spec */
#define NVME_MIN_ADMIN_ENTRIES (2)
#define NVME_MAX_ADMIN_ENTRIES (4096)
Preallocate a limited number of nvme_tracker objects per qpair, rather than dynamically creating them at runtime. Sponsored by: Intel
2012-10-18 00:44:39 +00:00
/*
* NVME_IO_ENTRIES defines the size of an I/O qpair's submission and completion
* queues, while NVME_IO_TRACKERS defines the maximum number of I/O that we
* will allow outstanding on an I/O qpair at any time. The only advantage in
* having IO_ENTRIES > IO_TRACKERS is for debugging purposes - when dumping
* the contents of the submission and completion queues, it will show a longer
* history of data.
*/
#define NVME_IO_ENTRIES (256)
#define NVME_IO_TRACKERS (128)
#define NVME_MIN_IO_TRACKERS (16)
#define NVME_MAX_IO_TRACKERS (1024)
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
/*
* NVME_MAX_IO_ENTRIES is not defined, since it is specified in CC.MQES
* for each controller.
*/
#define NVME_INT_COAL_TIME (0) /* disabled */
#define NVME_INT_COAL_THRESHOLD (0) /* 0-based */
#define NVME_MAX_NAMESPACES (16)
#define NVME_MAX_CONSUMERS (2)
Enable asynchronous event requests on non-Chatham devices. Also add logic to clean up all outstanding asynchronous event requests when resetting or shutting down the controller, since these requests will not be explicitly completed by the controller itself. Sponsored by: Intel
2013-03-26 18:37:36 +00:00
#define NVME_MAX_ASYNC_EVENTS (8)
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
Make nvme_ctrlr_reset a nop if a reset is already in progress. This protects against cases where a controller crashes with multiple I/O outstanding, each timing out and requesting controller resets simultaneously. While here, remove a debugging printf from a previous commit, and add more logging around I/O that need to be resubmitted after a controller reset. Sponsored by: Intel Reviewed by: carl
2013-03-26 20:56:58 +00:00
#define NVME_DEFAULT_TIMEOUT_PERIOD (30) /* in seconds */
Add a tunable for the I/O timeout interval. Default is still 30 seconds, but can be adjusted between a min/max of 5 and 120 seconds. Sponsored by: Intel Reviewed by: carl
2013-03-26 20:02:35 +00:00
#define NVME_MIN_TIMEOUT_PERIOD (5)
#define NVME_MAX_TIMEOUT_PERIOD (120)
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
Put the nvme_qpair mutex on its own cacheline. Sponsored by: Intel
2012-10-10 23:28:58 +00:00
#ifndef CACHE_LINE_SIZE
#define CACHE_LINE_SIZE (64)
#endif
Add struct nvme_request object which contains all of the parameters passed from an NVMe consumer. This allows us to mostly build NVMe command buffers without holding the qpair lock, and also allows for future queueing of nvme_request objects in cases where the submission queue is full and no nvme_tracker objects are available. Sponsored by: Intel
2012-10-18 00:38:28 +00:00
extern uma_zone_t nvme_request_zone;
struct nvme_request {
struct nvme_command cmd;
void *payload;
uint32_t payload_size;
Add a tunable for the I/O timeout interval. Default is still 30 seconds, but can be adjusted between a min/max of 5 and 120 seconds. Sponsored by: Intel Reviewed by: carl
2013-03-26 20:02:35 +00:00
boolean_t timeout;
Cleanup uio-related code to use struct nvme_request and nvme_ctrlr_submit_io_request(). While here, also fix case where a uio may have more than 1 iovec. NVMe's definition of SGEs (called PRPs) only allows for the first SGE to start on a non-page boundary. The simplest way to handle this is to construct a temporary uio for each iovec, and submit an NVMe request for each. Sponsored by: Intel
2012-10-18 00:40:40 +00:00
struct uio *uio;
Add struct nvme_request object which contains all of the parameters passed from an NVMe consumer. This allows us to mostly build NVMe command buffers without holding the qpair lock, and also allows for future queueing of nvme_request objects in cases where the submission queue is full and no nvme_tracker objects are available. Sponsored by: Intel
2012-10-18 00:38:28 +00:00
nvme_cb_fn_t cb_fn;
void *cb_arg;
Add ability to queue nvme_request objects if no nvme_trackers are available. This eliminates the need to manage queue depth at the nvd(4) level for Chatham prototype board workarounds, and also adds the ability to accept a number of requests on a single qpair that is much larger than the number of trackers allocated. Sponsored by: Intel
2012-10-18 00:45:53 +00:00
STAILQ_ENTRY(nvme_request) stailq;
Add struct nvme_request object which contains all of the parameters passed from an NVMe consumer. This allows us to mostly build NVMe command buffers without holding the qpair lock, and also allows for future queueing of nvme_request objects in cases where the submission queue is full and no nvme_tracker objects are available. Sponsored by: Intel
2012-10-18 00:38:28 +00:00
};
Enable asynchronous event requests on non-Chatham devices. Also add logic to clean up all outstanding asynchronous event requests when resetting or shutting down the controller, since these requests will not be explicitly completed by the controller itself. Sponsored by: Intel
2013-03-26 18:37:36 +00:00
struct nvme_async_event_request {
struct nvme_controller *ctrlr;
struct nvme_request *req;
};
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
struct nvme_tracker {
Keep a doubly-linked list of outstanding trackers. This enables in-order re-submission of I/O after a controller reset. Sponsored by: Intel
2013-03-26 18:45:16 +00:00
TAILQ_ENTRY(nvme_tracker) tailq;
Add struct nvme_request object which contains all of the parameters passed from an NVMe consumer. This allows us to mostly build NVMe command buffers without holding the qpair lock, and also allows for future queueing of nvme_request objects in cases where the submission queue is full and no nvme_tracker objects are available. Sponsored by: Intel
2012-10-18 00:38:28 +00:00
struct nvme_request *req;
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
struct nvme_qpair *qpair;
struct callout timer;
Merge struct nvme_prp_list into struct nvme_tracker. This simplifies the driver significantly where it is constructing commands to be submitted to hardware. By reducing the number of PRPs (NVMe parlance for SGE) from 128 to 32, it ensures we do not allocate too much memory for more common smaller I/O sizes, while still supporting up to 128KB I/O sizes. This also paves the way for pre-allocation of nvme_tracker objects for each queue which will simplify the I/O path even further. Sponsored by: Intel
2012-10-18 00:37:11 +00:00
bus_dmamap_t payload_dma_map;
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
uint16_t cid;
Merge struct nvme_prp_list into struct nvme_tracker. This simplifies the driver significantly where it is constructing commands to be submitted to hardware. By reducing the number of PRPs (NVMe parlance for SGE) from 128 to 32, it ensures we do not allocate too much memory for more common smaller I/O sizes, while still supporting up to 128KB I/O sizes. This also paves the way for pre-allocation of nvme_tracker objects for each queue which will simplify the I/O path even further. Sponsored by: Intel
2012-10-18 00:37:11 +00:00
uint64_t prp[NVME_MAX_PRP_LIST_ENTRIES];
bus_addr_t prp_bus_addr;
bus_dmamap_t prp_dma_map;
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
};
struct nvme_qpair {
struct nvme_controller *ctrlr;
uint32_t id;
uint32_t phase;
uint16_t vector;
int rid;
struct resource *res;
void *tag;
uint32_t max_xfer_size;
uint32_t num_entries;
Preallocate a limited number of nvme_tracker objects per qpair, rather than dynamically creating them at runtime. Sponsored by: Intel
2012-10-18 00:44:39 +00:00
uint32_t num_trackers;
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
uint32_t sq_tdbl_off;
uint32_t cq_hdbl_off;
uint32_t sq_head;
uint32_t sq_tail;
uint32_t cq_head;
int64_t num_cmds;
Count number of times each queue pair's interrupt handler is invoked. Also add sysctls to query and reset each queue pair's stats, including the new count added here. Sponsored by: Intel
2012-10-10 23:35:16 +00:00
int64_t num_intr_handler_calls;
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
struct nvme_command *cmd;
struct nvme_completion *cpl;
bus_dma_tag_t dma_tag;
bus_dmamap_t cmd_dma_map;
uint64_t cmd_bus_addr;
bus_dmamap_t cpl_dma_map;
uint64_t cpl_bus_addr;
Keep a doubly-linked list of outstanding trackers. This enables in-order re-submission of I/O after a controller reset. Sponsored by: Intel
2013-03-26 18:45:16 +00:00
TAILQ_HEAD(, nvme_tracker) free_tr;
TAILQ_HEAD(, nvme_tracker) outstanding_tr;
Add ability to queue nvme_request objects if no nvme_trackers are available. This eliminates the need to manage queue depth at the nvd(4) level for Chatham prototype board workarounds, and also adds the ability to accept a number of requests on a single qpair that is much larger than the number of trackers allocated. Sponsored by: Intel
2012-10-18 00:45:53 +00:00
STAILQ_HEAD(, nvme_request) queued_req;
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
struct nvme_tracker **act_tr;
Add controller reset capability to nvme(4) and ability to explicitly invoke it from nvmecontrol(8). Controller reset will be performed in cases where I/O are repeatedly timing out, the controller reports an unrecoverable condition, or when explicitly requested via IOCTL or an nvme consumer. Since the controller may be in such a state where it cannot even process queue deletion requests, we will perform a controller reset without trying to clean up anything on the controller first. Sponsored by: Intel Reviewed by: carl
2013-03-26 19:50:46 +00:00
boolean_t is_enabled;
Put the nvme_qpair mutex on its own cacheline. Sponsored by: Intel
2012-10-10 23:28:58 +00:00
struct mtx lock __aligned(CACHE_LINE_SIZE);
} __aligned(CACHE_LINE_SIZE);
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
struct nvme_namespace {
struct nvme_controller *ctrlr;
struct nvme_namespace_data data;
uint16_t id;
uint16_t flags;
struct cdev *cdev;
Add an interface for nvme shim drivers (i.e. nvd) to register for notifications when new nvme controllers are added to the system. Sponsored by: Intel
2013-03-26 18:39:54 +00:00
void *cons_cookie[NVME_MAX_CONSUMERS];
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
};
/*
* One of these per allocated PCI device.
*/
struct nvme_controller {
device_t dev;
uint32_t ready_timeout_in_ms;
bus_space_tag_t bus_tag;
bus_space_handle_t bus_handle;
int resource_id;
struct resource *resource;
Map BAR 4/5, because NVMe spec says devices may place the MSI-X table behind BAR 4/5, rather than in BAR 0/1 with the control/doorbell registers. Sponsored by: Intel
2012-12-18 23:27:18 +00:00
/*
* The NVMe spec allows for the MSI-X table to be placed in BAR 4/5,
* separate from the control registers which are in BAR 0/1. These
* members track the mapping of BAR 4/5 for that reason.
*/
int bar4_resource_id;
struct resource *bar4_resource;
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
#ifdef CHATHAM2
bus_space_tag_t chatham_bus_tag;
bus_space_handle_t chatham_bus_handle;
int chatham_resource_id;
struct resource *chatham_resource;
#endif
uint32_t msix_enabled;
uint32_t force_intx;
By default, always escalate to controller reset when an I/O times out. While aborts are typically cleaner than a full controller reset, many times an I/O timeout indicates other controller-level issues where aborts may not work. NVMe drivers for other operating systems are also defaulting to controller reset rather than aborts for timed out I/O. Sponsored by: Intel Reviewed by: carl
2013-03-26 20:32:57 +00:00
uint32_t enable_aborts;
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
uint32_t num_io_queues;
boolean_t per_cpu_io_queues;
/* Fields for tracking progress during controller initialization. */
struct intr_config_hook config_hook;
uint32_t ns_identified;
uint32_t queues_created;
Add controller reset capability to nvme(4) and ability to explicitly invoke it from nvmecontrol(8). Controller reset will be performed in cases where I/O are repeatedly timing out, the controller reports an unrecoverable condition, or when explicitly requested via IOCTL or an nvme consumer. Since the controller may be in such a state where it cannot even process queue deletion requests, we will perform a controller reset without trying to clean up anything on the controller first. Sponsored by: Intel Reviewed by: carl
2013-03-26 19:50:46 +00:00
uint32_t num_start_attempts;
By default, always escalate to controller reset when an I/O times out. While aborts are typically cleaner than a full controller reset, many times an I/O timeout indicates other controller-level issues where aborts may not work. NVMe drivers for other operating systems are also defaulting to controller reset rather than aborts for timed out I/O. Sponsored by: Intel Reviewed by: carl
2013-03-26 20:32:57 +00:00
struct task reset_task;
Add handling for controller fatal status (csts.cfs). On any I/O timeout, check for csts.cfs==1. If set, the controller is reporting fatal status and we reset the controller immediately, rather than trying to abort the timed out command. This changeset also includes deferring the controller start portion of the reset to a separate task. This ensures we are always performing a controller start operation from a consistent context. Sponsored by: Intel Reviewed by: carl
2013-03-26 19:58:17 +00:00
struct taskqueue *taskqueue;
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
/* For shared legacy interrupt. */
int rid;
struct resource *res;
void *tag;
bus_dma_tag_t hw_desc_tag;
bus_dmamap_t hw_desc_map;
/** maximum i/o size in bytes */
uint32_t max_xfer_size;
/** interrupt coalescing time period (in microseconds) */
uint32_t int_coal_time;
/** interrupt coalescing threshold */
uint32_t int_coal_threshold;
Add a tunable for the I/O timeout interval. Default is still 30 seconds, but can be adjusted between a min/max of 5 and 120 seconds. Sponsored by: Intel Reviewed by: carl
2013-03-26 20:02:35 +00:00
/** timeout period in seconds */
uint32_t timeout_period;
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
struct nvme_qpair adminq;
struct nvme_qpair *ioq;
struct nvme_registers *regs;
struct nvme_controller_data cdata;
struct nvme_namespace ns[NVME_MAX_NAMESPACES];
struct cdev *cdev;
boolean_t is_started;
Enable asynchronous event requests on non-Chatham devices. Also add logic to clean up all outstanding asynchronous event requests when resetting or shutting down the controller, since these requests will not be explicitly completed by the controller itself. Sponsored by: Intel
2013-03-26 18:37:36 +00:00
uint32_t num_aers;
struct nvme_async_event_request aer[NVME_MAX_ASYNC_EVENTS];
Add an interface for nvme shim drivers (i.e. nvd) to register for notifications when new nvme controllers are added to the system. Sponsored by: Intel
2013-03-26 18:39:54 +00:00
void *cons_cookie[NVME_MAX_CONSUMERS];
Make nvme_ctrlr_reset a nop if a reset is already in progress. This protects against cases where a controller crashes with multiple I/O outstanding, each timing out and requesting controller resets simultaneously. While here, remove a debugging printf from a previous commit, and add more logging around I/O that need to be resubmitted after a controller reset. Sponsored by: Intel Reviewed by: carl
2013-03-26 20:56:58 +00:00
uint32_t is_resetting;
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
#ifdef CHATHAM2
uint64_t chatham_size;
uint64_t chatham_lbas;
#endif
};
#define nvme_mmio_offsetof(reg) \
offsetof(struct nvme_registers, reg)
#define nvme_mmio_read_4(sc, reg) \
bus_space_read_4((sc)->bus_tag, (sc)->bus_handle, \
nvme_mmio_offsetof(reg))
#define nvme_mmio_write_4(sc, reg, val) \
bus_space_write_4((sc)->bus_tag, (sc)->bus_handle, \
nvme_mmio_offsetof(reg), val)
#define nvme_mmio_write_8(sc, reg, val) \
do { \
bus_space_write_4((sc)->bus_tag, (sc)->bus_handle, \
nvme_mmio_offsetof(reg), val & 0xFFFFFFFF); \
bus_space_write_4((sc)->bus_tag, (sc)->bus_handle, \
nvme_mmio_offsetof(reg)+4, \
(val & 0xFFFFFFFF00000000UL) >> 32); \
} while (0);
#ifdef CHATHAM2
#define chatham_read_4(softc, reg) \
bus_space_read_4((softc)->chatham_bus_tag, \
(softc)->chatham_bus_handle, reg)
#define chatham_write_8(sc, reg, val) \
do { \
bus_space_write_4((sc)->chatham_bus_tag, \
(sc)->chatham_bus_handle, reg, val & 0xffffffff); \
bus_space_write_4((sc)->chatham_bus_tag, \
(sc)->chatham_bus_handle, reg+4, \
(val & 0xFFFFFFFF00000000UL) >> 32); \
} while (0);
#endif /* CHATHAM2 */
#if __FreeBSD_version < 800054
#define wmb() __asm volatile("sfence" ::: "memory")
#define mb() __asm volatile("mfence" ::: "memory")
#endif
void nvme_ns_test(struct nvme_namespace *ns, u_long cmd, caddr_t arg);
void nvme_ctrlr_cmd_identify_controller(struct nvme_controller *ctrlr,
void *payload,
nvme_cb_fn_t cb_fn, void *cb_arg);
void nvme_ctrlr_cmd_identify_namespace(struct nvme_controller *ctrlr,
uint16_t nsid, void *payload,
nvme_cb_fn_t cb_fn, void *cb_arg);
void nvme_ctrlr_cmd_set_interrupt_coalescing(struct nvme_controller *ctrlr,
uint32_t microseconds,
uint32_t threshold,
nvme_cb_fn_t cb_fn,
void *cb_arg);
Add structure definitions and a controller command function for error log pages. Sponsored by: Intel Reviewed by: carl
2013-03-26 21:01:53 +00:00
void nvme_ctrlr_cmd_get_error_page(struct nvme_controller *ctrlr,
struct nvme_error_information_entry *payload,
uint32_t num_entries, /* 0 = max */
nvme_cb_fn_t cb_fn,
void *cb_arg);
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
void nvme_ctrlr_cmd_get_health_information_page(struct nvme_controller *ctrlr,
uint32_t nsid,
struct nvme_health_information_page *payload,
nvme_cb_fn_t cb_fn,
void *cb_arg);
Add structure definitions and controller command function for firmware log pages. Sponsored by: Intel Reviewed by: carl
2013-03-26 21:03:03 +00:00
void nvme_ctrlr_cmd_get_firmware_page(struct nvme_controller *ctrlr,
struct nvme_firmware_page *payload,
nvme_cb_fn_t cb_fn,
void *cb_arg);
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
void nvme_ctrlr_cmd_create_io_cq(struct nvme_controller *ctrlr,
struct nvme_qpair *io_que, uint16_t vector,
nvme_cb_fn_t cb_fn, void *cb_arg);
void nvme_ctrlr_cmd_create_io_sq(struct nvme_controller *ctrlr,
struct nvme_qpair *io_que,
nvme_cb_fn_t cb_fn, void *cb_arg);
void nvme_ctrlr_cmd_delete_io_cq(struct nvme_controller *ctrlr,
struct nvme_qpair *io_que,
nvme_cb_fn_t cb_fn, void *cb_arg);
void nvme_ctrlr_cmd_delete_io_sq(struct nvme_controller *ctrlr,
struct nvme_qpair *io_que,
nvme_cb_fn_t cb_fn, void *cb_arg);
void nvme_ctrlr_cmd_set_num_queues(struct nvme_controller *ctrlr,
uint32_t num_queues, nvme_cb_fn_t cb_fn,
void *cb_arg);
Enable asynchronous event requests on non-Chatham devices. Also add logic to clean up all outstanding asynchronous event requests when resetting or shutting down the controller, since these requests will not be explicitly completed by the controller itself. Sponsored by: Intel
2013-03-26 18:37:36 +00:00
void nvme_ctrlr_cmd_set_async_event_config(struct nvme_controller *ctrlr,
union nvme_critical_warning_state state,
nvme_cb_fn_t cb_fn, void *cb_arg);
Add support for ABORT commands, including issuing these commands when an I/O times out. Also ensure that we retry commands that are aborted due to a timeout. Sponsored by: Intel
2013-03-26 18:23:35 +00:00
void nvme_ctrlr_cmd_abort(struct nvme_controller *ctrlr, uint16_t cid,
uint16_t sqid, nvme_cb_fn_t cb_fn, void *cb_arg);
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
void nvme_payload_map(void *arg, bus_dma_segment_t *seg, int nseg,
int error);
Cleanup uio-related code to use struct nvme_request and nvme_ctrlr_submit_io_request(). While here, also fix case where a uio may have more than 1 iovec. NVMe's definition of SGEs (called PRPs) only allows for the first SGE to start on a non-page boundary. The simplest way to handle this is to construct a temporary uio for each iovec, and submit an NVMe request for each. Sponsored by: Intel
2012-10-18 00:40:40 +00:00
void nvme_payload_map_uio(void *arg, bus_dma_segment_t *seg, int nseg,
bus_size_t mapsize, int error);
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
int nvme_ctrlr_construct(struct nvme_controller *ctrlr, device_t dev);
Move controller destruction code from nvme_detach() to new nvme_ctrlr_destruct() function. Sponsored by: Intel
2013-03-26 18:34:19 +00:00
void nvme_ctrlr_destruct(struct nvme_controller *ctrlr, device_t dev);
Add controller reset capability to nvme(4) and ability to explicitly invoke it from nvmecontrol(8). Controller reset will be performed in cases where I/O are repeatedly timing out, the controller reports an unrecoverable condition, or when explicitly requested via IOCTL or an nvme consumer. Since the controller may be in such a state where it cannot even process queue deletion requests, we will perform a controller reset without trying to clean up anything on the controller first. Sponsored by: Intel Reviewed by: carl
2013-03-26 19:50:46 +00:00
int nvme_ctrlr_hw_reset(struct nvme_controller *ctrlr);
void nvme_ctrlr_reset(struct nvme_controller *ctrlr);
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
/* ctrlr defined as void * to allow use with config_intrhook. */
void nvme_ctrlr_start(void *ctrlr_arg);
Add nvme_ctrlr_submit_[admin|io]_request functions which consolidates code for allocating nvme_tracker objects and making calls into bus_dmamap_load for commands which have payloads. Sponsored by: Intel
2012-10-18 00:39:29 +00:00
void nvme_ctrlr_submit_admin_request(struct nvme_controller *ctrlr,
struct nvme_request *req);
void nvme_ctrlr_submit_io_request(struct nvme_controller *ctrlr,
struct nvme_request *req);
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
void nvme_qpair_construct(struct nvme_qpair *qpair, uint32_t id,
uint16_t vector, uint32_t num_entries,
Preallocate a limited number of nvme_tracker objects per qpair, rather than dynamically creating them at runtime. Sponsored by: Intel
2012-10-18 00:44:39 +00:00
uint32_t num_trackers, uint32_t max_xfer_size,
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
struct nvme_controller *ctrlr);
Add controller reset capability to nvme(4) and ability to explicitly invoke it from nvmecontrol(8). Controller reset will be performed in cases where I/O are repeatedly timing out, the controller reports an unrecoverable condition, or when explicitly requested via IOCTL or an nvme consumer. Since the controller may be in such a state where it cannot even process queue deletion requests, we will perform a controller reset without trying to clean up anything on the controller first. Sponsored by: Intel Reviewed by: carl
2013-03-26 19:50:46 +00:00
void nvme_qpair_submit_tracker(struct nvme_qpair *qpair,
struct nvme_tracker *tr);
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
void nvme_qpair_process_completions(struct nvme_qpair *qpair);
Create nvme_qpair_submit_request() which eliminates all of the code duplication between the admin and io controller-level submit functions. Sponsored by: Intel
2012-10-18 00:43:25 +00:00
void nvme_qpair_submit_request(struct nvme_qpair *qpair,
struct nvme_request *req);
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
Add controller reset capability to nvme(4) and ability to explicitly invoke it from nvmecontrol(8). Controller reset will be performed in cases where I/O are repeatedly timing out, the controller reports an unrecoverable condition, or when explicitly requested via IOCTL or an nvme consumer. Since the controller may be in such a state where it cannot even process queue deletion requests, we will perform a controller reset without trying to clean up anything on the controller first. Sponsored by: Intel Reviewed by: carl
2013-03-26 19:50:46 +00:00
void nvme_admin_qpair_enable(struct nvme_qpair *qpair);
void nvme_admin_qpair_disable(struct nvme_qpair *qpair);
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
void nvme_admin_qpair_destroy(struct nvme_qpair *qpair);
Add controller reset capability to nvme(4) and ability to explicitly invoke it from nvmecontrol(8). Controller reset will be performed in cases where I/O are repeatedly timing out, the controller reports an unrecoverable condition, or when explicitly requested via IOCTL or an nvme consumer. Since the controller may be in such a state where it cannot even process queue deletion requests, we will perform a controller reset without trying to clean up anything on the controller first. Sponsored by: Intel Reviewed by: carl
2013-03-26 19:50:46 +00:00
void nvme_io_qpair_enable(struct nvme_qpair *qpair);
void nvme_io_qpair_disable(struct nvme_qpair *qpair);
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
void nvme_io_qpair_destroy(struct nvme_qpair *qpair);
int nvme_ns_construct(struct nvme_namespace *ns, uint16_t id,
struct nvme_controller *ctrlr);
Add controller reset capability to nvme(4) and ability to explicitly invoke it from nvmecontrol(8). Controller reset will be performed in cases where I/O are repeatedly timing out, the controller reports an unrecoverable condition, or when explicitly requested via IOCTL or an nvme consumer. Since the controller may be in such a state where it cannot even process queue deletion requests, we will perform a controller reset without trying to clean up anything on the controller first. Sponsored by: Intel Reviewed by: carl
2013-03-26 19:50:46 +00:00
void nvme_ns_destruct(struct nvme_namespace *ns);
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
int nvme_ns_physio(struct cdev *dev, struct uio *uio, int ioflag);
void nvme_sysctl_initialize_ctrlr(struct nvme_controller *ctrlr);
void nvme_dump_command(struct nvme_command *cmd);
void nvme_dump_completion(struct nvme_completion *cpl);
static __inline void
nvme_single_map(void *arg, bus_dma_segment_t *seg, int nseg, int error)
{
uint64_t *bus_addr = (uint64_t *)arg;
*bus_addr = seg[0].ds_addr;
}
Add struct nvme_request object which contains all of the parameters passed from an NVMe consumer. This allows us to mostly build NVMe command buffers without holding the qpair lock, and also allows for future queueing of nvme_request objects in cases where the submission queue is full and no nvme_tracker objects are available. Sponsored by: Intel
2012-10-18 00:38:28 +00:00
static __inline struct nvme_request *
nvme_allocate_request(void *payload, uint32_t payload_size, nvme_cb_fn_t cb_fn,
void *cb_arg)
{
struct nvme_request *req;
req = uma_zalloc(nvme_request_zone, M_NOWAIT | M_ZERO);
if (req == NULL)
return (NULL);
req->payload = payload;
req->payload_size = payload_size;
req->cb_fn = cb_fn;
req->cb_arg = cb_arg;
Add a tunable for the I/O timeout interval. Default is still 30 seconds, but can be adjusted between a min/max of 5 and 120 seconds. Sponsored by: Intel Reviewed by: carl
2013-03-26 20:02:35 +00:00
req->timeout = TRUE;
Add struct nvme_request object which contains all of the parameters passed from an NVMe consumer. This allows us to mostly build NVMe command buffers without holding the qpair lock, and also allows for future queueing of nvme_request objects in cases where the submission queue is full and no nvme_tracker objects are available. Sponsored by: Intel
2012-10-18 00:38:28 +00:00
return (req);
}
Cleanup uio-related code to use struct nvme_request and nvme_ctrlr_submit_io_request(). While here, also fix case where a uio may have more than 1 iovec. NVMe's definition of SGEs (called PRPs) only allows for the first SGE to start on a non-page boundary. The simplest way to handle this is to construct a temporary uio for each iovec, and submit an NVMe request for each. Sponsored by: Intel
2012-10-18 00:40:40 +00:00
static __inline struct nvme_request *
nvme_allocate_request_uio(struct uio *uio, nvme_cb_fn_t cb_fn, void *cb_arg)
{
struct nvme_request *req;
req = uma_zalloc(nvme_request_zone, M_NOWAIT | M_ZERO);
if (req == NULL)
return (NULL);
req->uio = uio;
req->cb_fn = cb_fn;
req->cb_arg = cb_arg;
Add a tunable for the I/O timeout interval. Default is still 30 seconds, but can be adjusted between a min/max of 5 and 120 seconds. Sponsored by: Intel Reviewed by: carl
2013-03-26 20:02:35 +00:00
req->timeout = TRUE;
Cleanup uio-related code to use struct nvme_request and nvme_ctrlr_submit_io_request(). While here, also fix case where a uio may have more than 1 iovec. NVMe's definition of SGEs (called PRPs) only allows for the first SGE to start on a non-page boundary. The simplest way to handle this is to construct a temporary uio for each iovec, and submit an NVMe request for each. Sponsored by: Intel
2012-10-18 00:40:40 +00:00
return (req);
}
Add struct nvme_request object which contains all of the parameters passed from an NVMe consumer. This allows us to mostly build NVMe command buffers without holding the qpair lock, and also allows for future queueing of nvme_request objects in cases where the submission queue is full and no nvme_tracker objects are available. Sponsored by: Intel
2012-10-18 00:38:28 +00:00
#define nvme_free_request(req) uma_zfree(nvme_request_zone, req)
Add an interface for nvme shim drivers (i.e. nvd) to register for notifications when new nvme controllers are added to the system. Sponsored by: Intel
2013-03-26 18:39:54 +00:00
void nvme_notify_async_consumers(struct nvme_controller *ctrlr,
const struct nvme_completion *async_cpl);
This is the first of several commits which will add NVM Express (NVMe) support to FreeBSD. A full description of the overall functionality being added is below. nvmexpress.org defines NVM Express as "an optimized register interface, command set and feature set fo PCI Express (PCIe)-based Solid-State Drives (SSDs)." This commit adds nvme(4) and nvd(4) driver source code and Makefiles to the tree. Full NVMe functionality description: Add nvme(4) and nvd(4) drivers and nvmecontrol(8) for NVM Express (NVMe) device support. There will continue to be ongoing work on NVM Express support, but there is more than enough to allow for evaluation of pre-production NVM Express devices as well as soliciting feedback. Questions and feedback are welcome. nvme(4) implements NVMe hardware abstraction and is a provider of NVMe namespaces. The closest equivalent of an NVMe namespace is a SCSI LUN. nvd(4) is an NVMe consumer, surfacing NVMe namespaces as GEOM disks. nvmecontrol(8) is used for NVMe configuration and management. The following are currently supported: nvme(4) - full mandatory NVM command set support - per-CPU IO queues (enabled by default but configurable) - per-queue sysctls for statistics and full command/completion queue dumps for debugging - registration API for NVMe namespace consumers - I/O error handling (except for timeoutsee below) - compilation switches for support back to stable-7 nvd(4) - BIO_DELETE and BIO_FLUSH (if supported by controller) - proper BIO_ORDERED handling nvmecontrol(8) - devlist: list NVMe controllers and their namespaces - identify: display controller or namespace identify data in human-readable or hex format - perftest: quick and dirty performance test to measure raw performance of NVMe device without userspace/physio/GEOM overhead The following are still work in progress and will be completed over the next 3-6 months in rough priority order: - complete man pages - firmware download and activation - asynchronous error requests - command timeout error handling - controller resets - nvmecontrol(8) log page retrieval This has been primarily tested on amd64, with light testing on i386. I would be happy to provide assistance to anyone interested in porting this to other architectures, but am not currently planning to do this work myself. Big-endian and dmamap sync for command/completion queues are the main areas that would need to be addressed. The nvme(4) driver currently has references to Chatham, which is an Intel-developed prototype board which is not fully spec compliant. These references will all be removed over time. Sponsored by: Intel Contributions from: Joe Golio/EMC <joseph dot golio at emc dot com>
2012-09-17 19:23:01 +00:00
#endif /* __NVME_PRIVATE_H__ */
Reference in New Issue Copy Permalink