linux/Documentation/IPMI.txt
<<
>>
Prefs
   1
   2                          The Linux IPMI Driver
   3                          ---------------------
   4                              Corey Minyard
   5                          <minyard@mvista.com>
   6                            <minyard@acm.org>
   7
   8The Intelligent Platform Management Interface, or IPMI, is a
   9standard for controlling intelligent devices that monitor a system.
  10It provides for dynamic discovery of sensors in the system and the
  11ability to monitor the sensors and be informed when the sensor's
  12values change or go outside certain boundaries.  It also has a
  13standardized database for field-replaceable units (FRUs) and a watchdog
  14timer.
  15
  16To use this, you need an interface to an IPMI controller in your
  17system (called a Baseboard Management Controller, or BMC) and
  18management software that can use the IPMI system.
  19
  20This document describes how to use the IPMI driver for Linux.  If you
  21are not familiar with IPMI itself, see the web site at
  22http://www.intel.com/design/servers/ipmi/index.htm.  IPMI is a big
  23subject and I can't cover it all here!
  24
  25Configuration
  26-------------
  27
  28The Linux IPMI driver is modular, which means you have to pick several
  29things to have it work right depending on your hardware.  Most of
  30these are available in the 'Character Devices' menu then the IPMI
  31menu.
  32
  33No matter what, you must pick 'IPMI top-level message handler' to use
  34IPMI.  What you do beyond that depends on your needs and hardware.
  35
  36The message handler does not provide any user-level interfaces.
  37Kernel code (like the watchdog) can still use it.  If you need access
  38from userland, you need to select 'Device interface for IPMI' if you
  39want access through a device driver.
  40
  41The driver interface depends on your hardware.  If your system
  42properly provides the SMBIOS info for IPMI, the driver will detect it
  43and just work.  If you have a board with a standard interface (These
  44will generally be either "KCS", "SMIC", or "BT", consult your hardware
  45manual), choose the 'IPMI SI handler' option.  A driver also exists
  46for direct I2C access to the IPMI management controller.  Some boards
  47support this, but it is unknown if it will work on every board.  For
  48this, choose 'IPMI SMBus handler', but be ready to try to do some
  49figuring to see if it will work on your system if the SMBIOS/APCI
  50information is wrong or not present.  It is fairly safe to have both
  51these enabled and let the drivers auto-detect what is present.
  52
  53You should generally enable ACPI on your system, as systems with IPMI
  54can have ACPI tables describing them.
  55
  56If you have a standard interface and the board manufacturer has done
  57their job correctly, the IPMI controller should be automatically
  58detected (via ACPI or SMBIOS tables) and should just work.  Sadly,
  59many boards do not have this information.  The driver attempts
  60standard defaults, but they may not work.  If you fall into this
  61situation, you need to read the section below named 'The SI Driver' or
  62"The SMBus Driver" on how to hand-configure your system.
  63
  64IPMI defines a standard watchdog timer.  You can enable this with the
  65'IPMI Watchdog Timer' config option.  If you compile the driver into
  66the kernel, then via a kernel command-line option you can have the
  67watchdog timer start as soon as it initializes.  It also have a lot
  68of other options, see the 'Watchdog' section below for more details.
  69Note that you can also have the watchdog continue to run if it is
  70closed (by default it is disabled on close).  Go into the 'Watchdog
  71Cards' menu, enable 'Watchdog Timer Support', and enable the option
  72'Disable watchdog shutdown on close'.
  73
  74IPMI systems can often be powered off using IPMI commands.  Select
  75'IPMI Poweroff' to do this.  The driver will auto-detect if the system
  76can be powered off by IPMI.  It is safe to enable this even if your
  77system doesn't support this option.  This works on ATCA systems, the
  78Radisys CPI1 card, and any IPMI system that supports standard chassis
  79management commands.
  80
  81If you want the driver to put an event into the event log on a panic,
  82enable the 'Generate a panic event to all BMCs on a panic' option.  If
  83you want the whole panic string put into the event log using OEM
  84events, enable the 'Generate OEM events containing the panic string'
  85option.
  86
  87Basic Design
  88------------
  89
  90The Linux IPMI driver is designed to be very modular and flexible, you
  91only need to take the pieces you need and you can use it in many
  92different ways.  Because of that, it's broken into many chunks of
  93code.  These chunks (by module name) are:
  94
  95ipmi_msghandler - This is the central piece of software for the IPMI
  96system.  It handles all messages, message timing, and responses.  The
  97IPMI users tie into this, and the IPMI physical interfaces (called
  98System Management Interfaces, or SMIs) also tie in here.  This
  99provides the kernelland interface for IPMI, but does not provide an
 100interface for use by application processes.
 101
 102ipmi_devintf - This provides a userland IOCTL interface for the IPMI
 103driver, each open file for this device ties in to the message handler
 104as an IPMI user.
 105
 106ipmi_si - A driver for various system interfaces.  This supports KCS,
 107SMIC, and BT interfaces.  Unless you have an SMBus interface or your
 108own custom interface, you probably need to use this.
 109
 110ipmi_smb - A driver for accessing BMCs on the SMBus. It uses the
 111I2C kernel driver's SMBus interfaces to send and receive IPMI messages
 112over the SMBus.
 113
 114ipmi_watchdog - IPMI requires systems to have a very capable watchdog
 115timer.  This driver implements the standard Linux watchdog timer
 116interface on top of the IPMI message handler.
 117
 118ipmi_poweroff - Some systems support the ability to be turned off via
 119IPMI commands.
 120
 121These are all individually selectable via configuration options.
 122
 123Note that the KCS-only interface has been removed.  The af_ipmi driver
 124is no longer supported and has been removed because it was impossible
 125to do 32 bit emulation on 64-bit kernels with it.
 126
 127Much documentation for the interface is in the include files.  The
 128IPMI include files are:
 129
 130net/af_ipmi.h - Contains the socket interface.
 131
 132linux/ipmi.h - Contains the user interface and IOCTL interface for IPMI.
 133
 134linux/ipmi_smi.h - Contains the interface for system management interfaces
 135(things that interface to IPMI controllers) to use.
 136
 137linux/ipmi_msgdefs.h - General definitions for base IPMI messaging.
 138
 139
 140Addressing
 141----------
 142
 143The IPMI addressing works much like IP addresses, you have an overlay
 144to handle the different address types.  The overlay is:
 145
 146  struct ipmi_addr
 147  {
 148        int   addr_type;
 149        short channel;
 150        char  data[IPMI_MAX_ADDR_SIZE];
 151  };
 152
 153The addr_type determines what the address really is.  The driver
 154currently understands two different types of addresses.
 155
 156"System Interface" addresses are defined as:
 157
 158  struct ipmi_system_interface_addr
 159  {
 160        int   addr_type;
 161        short channel;
 162  };
 163
 164and the type is IPMI_SYSTEM_INTERFACE_ADDR_TYPE.  This is used for talking
 165straight to the BMC on the current card.  The channel must be
 166IPMI_BMC_CHANNEL.
 167
 168Messages that are destined to go out on the IPMB bus use the
 169IPMI_IPMB_ADDR_TYPE address type.  The format is
 170
 171  struct ipmi_ipmb_addr
 172  {
 173        int           addr_type;
 174        short         channel;
 175        unsigned char slave_addr;
 176        unsigned char lun;
 177  };
 178
 179The "channel" here is generally zero, but some devices support more
 180than one channel, it corresponds to the channel as defined in the IPMI
 181spec.
 182
 183
 184Messages
 185--------
 186
 187Messages are defined as:
 188
 189struct ipmi_msg
 190{
 191        unsigned char netfn;
 192        unsigned char lun;
 193        unsigned char cmd;
 194        unsigned char *data;
 195        int           data_len;
 196};
 197
 198The driver takes care of adding/stripping the header information.  The
 199data portion is just the data to be send (do NOT put addressing info
 200here) or the response.  Note that the completion code of a response is
 201the first item in "data", it is not stripped out because that is how
 202all the messages are defined in the spec (and thus makes counting the
 203offsets a little easier :-).
 204
 205When using the IOCTL interface from userland, you must provide a block
 206of data for "data", fill it, and set data_len to the length of the
 207block of data, even when receiving messages.  Otherwise the driver
 208will have no place to put the message.
 209
 210Messages coming up from the message handler in kernelland will come in
 211as:
 212
 213  struct ipmi_recv_msg
 214  {
 215        struct list_head link;
 216
 217        /* The type of message as defined in the "Receive Types"
 218           defines above. */
 219        int         recv_type;
 220
 221        ipmi_user_t      *user;
 222        struct ipmi_addr addr;
 223        long             msgid;
 224        struct ipmi_msg  msg;
 225
 226        /* Call this when done with the message.  It will presumably free
 227           the message and do any other necessary cleanup. */
 228        void (*done)(struct ipmi_recv_msg *msg);
 229
 230        /* Place-holder for the data, don't make any assumptions about
 231           the size or existence of this, since it may change. */
 232        unsigned char   msg_data[IPMI_MAX_MSG_LENGTH];
 233  };
 234
 235You should look at the receive type and handle the message
 236appropriately.
 237
 238
 239The Upper Layer Interface (Message Handler)
 240-------------------------------------------
 241
 242The upper layer of the interface provides the users with a consistent
 243view of the IPMI interfaces.  It allows multiple SMI interfaces to be
 244addressed (because some boards actually have multiple BMCs on them)
 245and the user should not have to care what type of SMI is below them.
 246
 247
 248Creating the User
 249
 250To user the message handler, you must first create a user using
 251ipmi_create_user.  The interface number specifies which SMI you want
 252to connect to, and you must supply callback functions to be called
 253when data comes in.  The callback function can run at interrupt level,
 254so be careful using the callbacks.  This also allows to you pass in a
 255piece of data, the handler_data, that will be passed back to you on
 256all calls.
 257
 258Once you are done, call ipmi_destroy_user() to get rid of the user.
 259
 260From userland, opening the device automatically creates a user, and
 261closing the device automatically destroys the user.
 262
 263
 264Messaging
 265
 266To send a message from kernel-land, the ipmi_request() call does
 267pretty much all message handling.  Most of the parameter are
 268self-explanatory.  However, it takes a "msgid" parameter.  This is NOT
 269the sequence number of messages.  It is simply a long value that is
 270passed back when the response for the message is returned.  You may
 271use it for anything you like.
 272
 273Responses come back in the function pointed to by the ipmi_recv_hndl
 274field of the "handler" that you passed in to ipmi_create_user().
 275Remember again, these may be running at interrupt level.  Remember to
 276look at the receive type, too.
 277
 278From userland, you fill out an ipmi_req_t structure and use the
 279IPMICTL_SEND_COMMAND ioctl.  For incoming stuff, you can use select()
 280or poll() to wait for messages to come in.  However, you cannot use
 281read() to get them, you must call the IPMICTL_RECEIVE_MSG with the
 282ipmi_recv_t structure to actually get the message.  Remember that you
 283must supply a pointer to a block of data in the msg.data field, and
 284you must fill in the msg.data_len field with the size of the data.
 285This gives the receiver a place to actually put the message.
 286
 287If the message cannot fit into the data you provide, you will get an
 288EMSGSIZE error and the driver will leave the data in the receive
 289queue.  If you want to get it and have it truncate the message, us
 290the IPMICTL_RECEIVE_MSG_TRUNC ioctl.
 291
 292When you send a command (which is defined by the lowest-order bit of
 293the netfn per the IPMI spec) on the IPMB bus, the driver will
 294automatically assign the sequence number to the command and save the
 295command.  If the response is not receive in the IPMI-specified 5
 296seconds, it will generate a response automatically saying the command
 297timed out.  If an unsolicited response comes in (if it was after 5
 298seconds, for instance), that response will be ignored.
 299
 300In kernelland, after you receive a message and are done with it, you
 301MUST call ipmi_free_recv_msg() on it, or you will leak messages.  Note
 302that you should NEVER mess with the "done" field of a message, that is
 303required to properly clean up the message.
 304
 305Note that when sending, there is an ipmi_request_supply_msgs() call
 306that lets you supply the smi and receive message.  This is useful for
 307pieces of code that need to work even if the system is out of buffers
 308(the watchdog timer uses this, for instance).  You supply your own
 309buffer and own free routines.  This is not recommended for normal use,
 310though, since it is tricky to manage your own buffers.
 311
 312
 313Events and Incoming Commands
 314
 315The driver takes care of polling for IPMI events and receiving
 316commands (commands are messages that are not responses, they are
 317commands that other things on the IPMB bus have sent you).  To receive
 318these, you must register for them, they will not automatically be sent
 319to you.
 320
 321To receive events, you must call ipmi_set_gets_events() and set the
 322"val" to non-zero.  Any events that have been received by the driver
 323since startup will immediately be delivered to the first user that
 324registers for events.  After that, if multiple users are registered
 325for events, they will all receive all events that come in.
 326
 327For receiving commands, you have to individually register commands you
 328want to receive.  Call ipmi_register_for_cmd() and supply the netfn
 329and command name for each command you want to receive.  You also
 330specify a bitmask of the channels you want to receive the command from
 331(or use IPMI_CHAN_ALL for all channels if you don't care).  Only one
 332user may be registered for each netfn/cmd/channel, but different users
 333may register for different commands, or the same command if the
 334channel bitmasks do not overlap.
 335
 336From userland, equivalent IOCTLs are provided to do these functions.
 337
 338
 339The Lower Layer (SMI) Interface
 340-------------------------------
 341
 342As mentioned before, multiple SMI interfaces may be registered to the
 343message handler, each of these is assigned an interface number when
 344they register with the message handler.  They are generally assigned
 345in the order they register, although if an SMI unregisters and then
 346another one registers, all bets are off.
 347
 348The ipmi_smi.h defines the interface for management interfaces, see
 349that for more details.
 350
 351
 352The SI Driver
 353-------------
 354
 355The SI driver allows up to 4 KCS or SMIC interfaces to be configured
 356in the system.  By default, scan the ACPI tables for interfaces, and
 357if it doesn't find any the driver will attempt to register one KCS
 358interface at the spec-specified I/O port 0xca2 without interrupts.
 359You can change this at module load time (for a module) with:
 360
 361  modprobe ipmi_si.o type=<type1>,<type2>....
 362       ports=<port1>,<port2>... addrs=<addr1>,<addr2>...
 363       irqs=<irq1>,<irq2>... trydefaults=[0|1]
 364       regspacings=<sp1>,<sp2>,... regsizes=<size1>,<size2>,...
 365       regshifts=<shift1>,<shift2>,...
 366       slave_addrs=<addr1>,<addr2>,...
 367       force_kipmid=<enable1>,<enable2>,...
 368       kipmid_max_busy_us=<ustime1>,<ustime2>,...
 369       unload_when_empty=[0|1]
 370
 371Each of these except si_trydefaults is a list, the first item for the
 372first interface, second item for the second interface, etc.
 373
 374The si_type may be either "kcs", "smic", or "bt".  If you leave it blank, it
 375defaults to "kcs".
 376
 377If you specify si_addrs as non-zero for an interface, the driver will
 378use the memory address given as the address of the device.  This
 379overrides si_ports.
 380
 381If you specify si_ports as non-zero for an interface, the driver will
 382use the I/O port given as the device address.
 383
 384If you specify si_irqs as non-zero for an interface, the driver will
 385attempt to use the given interrupt for the device.
 386
 387si_trydefaults sets whether the standard IPMI interface at 0xca2 and
 388any interfaces specified by ACPE are tried.  By default, the driver
 389tries it, set this value to zero to turn this off.
 390
 391The next three parameters have to do with register layout.  The
 392registers used by the interfaces may not appear at successive
 393locations and they may not be in 8-bit registers.  These parameters
 394allow the layout of the data in the registers to be more precisely
 395specified.
 396
 397The regspacings parameter give the number of bytes between successive
 398register start addresses.  For instance, if the regspacing is set to 4
 399and the start address is 0xca2, then the address for the second
 400register would be 0xca6.  This defaults to 1.
 401
 402The regsizes parameter gives the size of a register, in bytes.  The
 403data used by IPMI is 8-bits wide, but it may be inside a larger
 404register.  This parameter allows the read and write type to specified.
 405It may be 1, 2, 4, or 8.  The default is 1.
 406
 407Since the register size may be larger than 32 bits, the IPMI data may not
 408be in the lower 8 bits.  The regshifts parameter give the amount to shift
 409the data to get to the actual IPMI data.
 410
 411The slave_addrs specifies the IPMI address of the local BMC.  This is
 412usually 0x20 and the driver defaults to that, but in case it's not, it
 413can be specified when the driver starts up.
 414
 415The force_ipmid parameter forcefully enables (if set to 1) or disables
 416(if set to 0) the kernel IPMI daemon.  Normally this is auto-detected
 417by the driver, but systems with broken interrupts might need an enable,
 418or users that don't want the daemon (don't need the performance, don't
 419want the CPU hit) can disable it.
 420
 421If unload_when_empty is set to 1, the driver will be unloaded if it
 422doesn't find any interfaces or all the interfaces fail to work.  The
 423default is one.  Setting to 0 is useful with the hotmod, but is
 424obviously only useful for modules.
 425
 426When compiled into the kernel, the parameters can be specified on the
 427kernel command line as:
 428
 429  ipmi_si.type=<type1>,<type2>...
 430       ipmi_si.ports=<port1>,<port2>... ipmi_si.addrs=<addr1>,<addr2>...
 431       ipmi_si.irqs=<irq1>,<irq2>... ipmi_si.trydefaults=[0|1]
 432       ipmi_si.regspacings=<sp1>,<sp2>,...
 433       ipmi_si.regsizes=<size1>,<size2>,...
 434       ipmi_si.regshifts=<shift1>,<shift2>,...
 435       ipmi_si.slave_addrs=<addr1>,<addr2>,...
 436       ipmi_si.force_kipmid=<enable1>,<enable2>,...
 437       ipmi_si.kipmid_max_busy_us=<ustime1>,<ustime2>,...
 438
 439It works the same as the module parameters of the same names.
 440
 441By default, the driver will attempt to detect any device specified by
 442ACPI, and if none of those then a KCS device at the spec-specified
 4430xca2.  If you want to turn this off, set the "trydefaults" option to
 444false.
 445
 446If your IPMI interface does not support interrupts and is a KCS or
 447SMIC interface, the IPMI driver will start a kernel thread for the
 448interface to help speed things up.  This is a low-priority kernel
 449thread that constantly polls the IPMI driver while an IPMI operation
 450is in progress.  The force_kipmid module parameter will all the user to
 451force this thread on or off.  If you force it off and don't have
 452interrupts, the driver will run VERY slowly.  Don't blame me,
 453these interfaces suck.
 454
 455Unfortunately, this thread can use a lot of CPU depending on the
 456interface's performance.  This can waste a lot of CPU and cause
 457various issues with detecting idle CPU and using extra power.  To
 458avoid this, the kipmid_max_busy_us sets the maximum amount of time, in
 459microseconds, that kipmid will spin before sleeping for a tick.  This
 460value sets a balance between performance and CPU waste and needs to be
 461tuned to your needs.  Maybe, someday, auto-tuning will be added, but
 462that's not a simple thing and even the auto-tuning would need to be
 463tuned to the user's desired performance.
 464
 465The driver supports a hot add and remove of interfaces.  This way,
 466interfaces can be added or removed after the kernel is up and running.
 467This is done using /sys/modules/ipmi_si/parameters/hotmod, which is a
 468write-only parameter.  You write a string to this interface.  The string
 469has the format:
 470   <op1>[:op2[:op3...]]
 471The "op"s are:
 472   add|remove,kcs|bt|smic,mem|i/o,<address>[,<opt1>[,<opt2>[,...]]]
 473You can specify more than one interface on the line.  The "opt"s are:
 474   rsp=<regspacing>
 475   rsi=<regsize>
 476   rsh=<regshift>
 477   irq=<irq>
 478   ipmb=<ipmb slave addr>
 479and these have the same meanings as discussed above.  Note that you
 480can also use this on the kernel command line for a more compact format
 481for specifying an interface.  Note that when removing an interface,
 482only the first three parameters (si type, address type, and address)
 483are used for the comparison.  Any options are ignored for removing.
 484
 485The SMBus Driver
 486----------------
 487
 488The SMBus driver allows up to 4 SMBus devices to be configured in the
 489system.  By default, the driver will register any SMBus interfaces it finds
 490in the I2C address range of 0x20 to 0x4f on any adapter.  You can change this
 491at module load time (for a module) with:
 492
 493  modprobe ipmi_smb.o
 494        addr=<adapter1>,<i2caddr1>[,<adapter2>,<i2caddr2>[,...]]
 495        dbg=<flags1>,<flags2>...
 496        [defaultprobe=1] [dbg_probe=1]
 497
 498The addresses are specified in pairs, the first is the adapter ID and the
 499second is the I2C address on that adapter.
 500
 501The debug flags are bit flags for each BMC found, they are:
 502IPMI messages: 1, driver state: 2, timing: 4, I2C probe: 8
 503
 504Setting smb_defaultprobe to zero disabled the default probing of SMBus
 505interfaces at address range 0x20 to 0x4f.  This means that only the
 506BMCs specified on the smb_addr line will be detected.
 507
 508Setting smb_dbg_probe to 1 will enable debugging of the probing and
 509detection process for BMCs on the SMBusses.
 510
 511Discovering the IPMI compliant BMC on the SMBus can cause devices
 512on the I2C bus to fail. The SMBus driver writes a "Get Device ID" IPMI
 513message as a block write to the I2C bus and waits for a response.
 514This action can be detrimental to some I2C devices. It is highly recommended
 515that the known I2c address be given to the SMBus driver in the smb_addr
 516parameter. The default address range will not be used when a smb_addr
 517parameter is provided.
 518
 519When compiled into the kernel, the addresses can be specified on the
 520kernel command line as:
 521
 522  ipmb_smb.addr=<adapter1>,<i2caddr1>[,<adapter2>,<i2caddr2>[,...]]
 523        ipmi_smb.dbg=<flags1>,<flags2>...
 524        ipmi_smb.defaultprobe=0 ipmi_smb.dbg_probe=1
 525
 526These are the same options as on the module command line.
 527
 528Note that you might need some I2C changes if CONFIG_IPMI_PANIC_EVENT
 529is enabled along with this, so the I2C driver knows to run to
 530completion during sending a panic event.
 531
 532
 533Other Pieces
 534------------
 535
 536Get the detailed info related with the IPMI device
 537--------------------------------------------------
 538
 539Some users need more detailed information about a device, like where
 540the address came from or the raw base device for the IPMI interface.
 541You can use the IPMI smi_watcher to catch the IPMI interfaces as they
 542come or go, and to grab the information, you can use the function
 543ipmi_get_smi_info(), which returns the following structure:
 544
 545struct ipmi_smi_info {
 546        enum ipmi_addr_src addr_src;
 547        struct device *dev;
 548        union {
 549                struct {
 550                        void *acpi_handle;
 551                } acpi_info;
 552        } addr_info;
 553};
 554
 555Currently special info for only for SI_ACPI address sources is
 556returned.  Others may be added as necessary.
 557
 558Note that the dev pointer is included in the above structure, and
 559assuming ipmi_smi_get_info returns success, you must call put_device
 560on the dev pointer.
 561
 562
 563Watchdog
 564--------
 565
 566A watchdog timer is provided that implements the Linux-standard
 567watchdog timer interface.  It has three module parameters that can be
 568used to control it:
 569
 570  modprobe ipmi_watchdog timeout=<t> pretimeout=<t> action=<action type>
 571      preaction=<preaction type> preop=<preop type> start_now=x
 572      nowayout=x ifnum_to_use=n
 573
 574ifnum_to_use specifies which interface the watchdog timer should use.
 575The default is -1, which means to pick the first one registered.
 576
 577The timeout is the number of seconds to the action, and the pretimeout
 578is the amount of seconds before the reset that the pre-timeout panic will
 579occur (if pretimeout is zero, then pretimeout will not be enabled).  Note
 580that the pretimeout is the time before the final timeout.  So if the
 581timeout is 50 seconds and the pretimeout is 10 seconds, then the pretimeout
 582will occur in 40 second (10 seconds before the timeout).
 583
 584The action may be "reset", "power_cycle", or "power_off", and
 585specifies what to do when the timer times out, and defaults to
 586"reset".
 587
 588The preaction may be "pre_smi" for an indication through the SMI
 589interface, "pre_int" for an indication through the SMI with an
 590interrupts, and "pre_nmi" for a NMI on a preaction.  This is how
 591the driver is informed of the pretimeout.
 592
 593The preop may be set to "preop_none" for no operation on a pretimeout,
 594"preop_panic" to set the preoperation to panic, or "preop_give_data"
 595to provide data to read from the watchdog device when the pretimeout
 596occurs.  A "pre_nmi" setting CANNOT be used with "preop_give_data"
 597because you can't do data operations from an NMI.
 598
 599When preop is set to "preop_give_data", one byte comes ready to read
 600on the device when the pretimeout occurs.  Select and fasync work on
 601the device, as well.
 602
 603If start_now is set to 1, the watchdog timer will start running as
 604soon as the driver is loaded.
 605
 606If nowayout is set to 1, the watchdog timer will not stop when the
 607watchdog device is closed.  The default value of nowayout is true
 608if the CONFIG_WATCHDOG_NOWAYOUT option is enabled, or false if not.
 609
 610When compiled into the kernel, the kernel command line is available
 611for configuring the watchdog:
 612
 613  ipmi_watchdog.timeout=<t> ipmi_watchdog.pretimeout=<t>
 614        ipmi_watchdog.action=<action type>
 615        ipmi_watchdog.preaction=<preaction type>
 616        ipmi_watchdog.preop=<preop type>
 617        ipmi_watchdog.start_now=x
 618        ipmi_watchdog.nowayout=x
 619
 620The options are the same as the module parameter options.
 621
 622The watchdog will panic and start a 120 second reset timeout if it
 623gets a pre-action.  During a panic or a reboot, the watchdog will
 624start a 120 timer if it is running to make sure the reboot occurs.
 625
 626Note that if you use the NMI preaction for the watchdog, you MUST NOT
 627use the nmi watchdog.  There is no reasonable way to tell if an NMI
 628comes from the IPMI controller, so it must assume that if it gets an
 629otherwise unhandled NMI, it must be from IPMI and it will panic
 630immediately.
 631
 632Once you open the watchdog timer, you must write a 'V' character to the
 633device to close it, or the timer will not stop.  This is a new semantic
 634for the driver, but makes it consistent with the rest of the watchdog
 635drivers in Linux.
 636
 637
 638Panic Timeouts
 639--------------
 640
 641The OpenIPMI driver supports the ability to put semi-custom and custom
 642events in the system event log if a panic occurs.  if you enable the
 643'Generate a panic event to all BMCs on a panic' option, you will get
 644one event on a panic in a standard IPMI event format.  If you enable
 645the 'Generate OEM events containing the panic string' option, you will
 646also get a bunch of OEM events holding the panic string.
 647
 648
 649The field settings of the events are:
 650* Generator ID: 0x21 (kernel)
 651* EvM Rev: 0x03 (this event is formatting in IPMI 1.0 format)
 652* Sensor Type: 0x20 (OS critical stop sensor)
 653* Sensor #: The first byte of the panic string (0 if no panic string)
 654* Event Dir | Event Type: 0x6f (Assertion, sensor-specific event info)
 655* Event Data 1: 0xa1 (Runtime stop in OEM bytes 2 and 3)
 656* Event data 2: second byte of panic string
 657* Event data 3: third byte of panic string
 658See the IPMI spec for the details of the event layout.  This event is
 659always sent to the local management controller.  It will handle routing
 660the message to the right place
 661
 662Other OEM events have the following format:
 663Record ID (bytes 0-1): Set by the SEL.
 664Record type (byte 2): 0xf0 (OEM non-timestamped)
 665byte 3: The slave address of the card saving the panic
 666byte 4: A sequence number (starting at zero)
 667The rest of the bytes (11 bytes) are the panic string.  If the panic string
 668is longer than 11 bytes, multiple messages will be sent with increasing
 669sequence numbers.
 670
 671Because you cannot send OEM events using the standard interface, this
 672function will attempt to find an SEL and add the events there.  It
 673will first query the capabilities of the local management controller.
 674If it has an SEL, then they will be stored in the SEL of the local
 675management controller.  If not, and the local management controller is
 676an event generator, the event receiver from the local management
 677controller will be queried and the events sent to the SEL on that
 678device.  Otherwise, the events go nowhere since there is nowhere to
 679send them.
 680
 681
 682Poweroff
 683--------
 684
 685If the poweroff capability is selected, the IPMI driver will install
 686a shutdown function into the standard poweroff function pointer.  This
 687is in the ipmi_poweroff module.  When the system requests a powerdown,
 688it will send the proper IPMI commands to do this.  This is supported on
 689several platforms.
 690
 691There is a module parameter named "poweroff_powercycle" that may
 692either be zero (do a power down) or non-zero (do a power cycle, power
 693the system off, then power it on in a few seconds).  Setting
 694ipmi_poweroff.poweroff_control=x will do the same thing on the kernel
 695command line.  The parameter is also available via the proc filesystem
 696in /proc/sys/dev/ipmi/poweroff_powercycle.  Note that if the system
 697does not support power cycling, it will always do the power off.
 698
 699The "ifnum_to_use" parameter specifies which interface the poweroff
 700code should use.  The default is -1, which means to pick the first one
 701registered.
 702
 703Note that if you have ACPI enabled, the system will prefer using ACPI to
 704power off.
 705