Back to home page

OSCL-LXR

 
 

    


0001 =====================
0002 The Linux IPMI Driver
0003 =====================
0004 
0005 :Author: Corey Minyard <minyard@mvista.com> / <minyard@acm.org>
0006 
0007 The Intelligent Platform Management Interface, or IPMI, is a
0008 standard for controlling intelligent devices that monitor a system.
0009 It provides for dynamic discovery of sensors in the system and the
0010 ability to monitor the sensors and be informed when the sensor's
0011 values change or go outside certain boundaries.  It also has a
0012 standardized database for field-replaceable units (FRUs) and a watchdog
0013 timer.
0014 
0015 To use this, you need an interface to an IPMI controller in your
0016 system (called a Baseboard Management Controller, or BMC) and
0017 management software that can use the IPMI system.
0018 
0019 This document describes how to use the IPMI driver for Linux.  If you
0020 are not familiar with IPMI itself, see the web site at
0021 https://www.intel.com/design/servers/ipmi/index.htm.  IPMI is a big
0022 subject and I can't cover it all here!
0023 
0024 Configuration
0025 -------------
0026 
0027 The Linux IPMI driver is modular, which means you have to pick several
0028 things to have it work right depending on your hardware.  Most of
0029 these are available in the 'Character Devices' menu then the IPMI
0030 menu.
0031 
0032 No matter what, you must pick 'IPMI top-level message handler' to use
0033 IPMI.  What you do beyond that depends on your needs and hardware.
0034 
0035 The message handler does not provide any user-level interfaces.
0036 Kernel code (like the watchdog) can still use it.  If you need access
0037 from userland, you need to select 'Device interface for IPMI' if you
0038 want access through a device driver.
0039 
0040 The driver interface depends on your hardware.  If your system
0041 properly provides the SMBIOS info for IPMI, the driver will detect it
0042 and just work.  If you have a board with a standard interface (These
0043 will generally be either "KCS", "SMIC", or "BT", consult your hardware
0044 manual), choose the 'IPMI SI handler' option.  A driver also exists
0045 for direct I2C access to the IPMI management controller.  Some boards
0046 support this, but it is unknown if it will work on every board.  For
0047 this, choose 'IPMI SMBus handler', but be ready to try to do some
0048 figuring to see if it will work on your system if the SMBIOS/APCI
0049 information is wrong or not present.  It is fairly safe to have both
0050 these enabled and let the drivers auto-detect what is present.
0051 
0052 You should generally enable ACPI on your system, as systems with IPMI
0053 can have ACPI tables describing them.
0054 
0055 If you have a standard interface and the board manufacturer has done
0056 their job correctly, the IPMI controller should be automatically
0057 detected (via ACPI or SMBIOS tables) and should just work.  Sadly,
0058 many boards do not have this information.  The driver attempts
0059 standard defaults, but they may not work.  If you fall into this
0060 situation, you need to read the section below named 'The SI Driver' or
0061 "The SMBus Driver" on how to hand-configure your system.
0062 
0063 IPMI defines a standard watchdog timer.  You can enable this with the
0064 'IPMI Watchdog Timer' config option.  If you compile the driver into
0065 the kernel, then via a kernel command-line option you can have the
0066 watchdog timer start as soon as it initializes.  It also have a lot
0067 of other options, see the 'Watchdog' section below for more details.
0068 Note that you can also have the watchdog continue to run if it is
0069 closed (by default it is disabled on close).  Go into the 'Watchdog
0070 Cards' menu, enable 'Watchdog Timer Support', and enable the option
0071 'Disable watchdog shutdown on close'.
0072 
0073 IPMI systems can often be powered off using IPMI commands.  Select
0074 'IPMI Poweroff' to do this.  The driver will auto-detect if the system
0075 can be powered off by IPMI.  It is safe to enable this even if your
0076 system doesn't support this option.  This works on ATCA systems, the
0077 Radisys CPI1 card, and any IPMI system that supports standard chassis
0078 management commands.
0079 
0080 If you want the driver to put an event into the event log on a panic,
0081 enable the 'Generate a panic event to all BMCs on a panic' option.  If
0082 you want the whole panic string put into the event log using OEM
0083 events, enable the 'Generate OEM events containing the panic string'
0084 option.  You can also enable these dynamically by setting the module
0085 parameter named "panic_op" in the ipmi_msghandler module to "event"
0086 or "string".  Setting that parameter to "none" disables this function.
0087 
0088 Basic Design
0089 ------------
0090 
0091 The Linux IPMI driver is designed to be very modular and flexible, you
0092 only need to take the pieces you need and you can use it in many
0093 different ways.  Because of that, it's broken into many chunks of
0094 code.  These chunks (by module name) are:
0095 
0096 ipmi_msghandler - This is the central piece of software for the IPMI
0097 system.  It handles all messages, message timing, and responses.  The
0098 IPMI users tie into this, and the IPMI physical interfaces (called
0099 System Management Interfaces, or SMIs) also tie in here.  This
0100 provides the kernelland interface for IPMI, but does not provide an
0101 interface for use by application processes.
0102 
0103 ipmi_devintf - This provides a userland IOCTL interface for the IPMI
0104 driver, each open file for this device ties in to the message handler
0105 as an IPMI user.
0106 
0107 ipmi_si - A driver for various system interfaces.  This supports KCS,
0108 SMIC, and BT interfaces.  Unless you have an SMBus interface or your
0109 own custom interface, you probably need to use this.
0110 
0111 ipmi_ssif - A driver for accessing BMCs on the SMBus. It uses the
0112 I2C kernel driver's SMBus interfaces to send and receive IPMI messages
0113 over the SMBus.
0114 
0115 ipmi_powernv - A driver for access BMCs on POWERNV systems.
0116 
0117 ipmi_watchdog - IPMI requires systems to have a very capable watchdog
0118 timer.  This driver implements the standard Linux watchdog timer
0119 interface on top of the IPMI message handler.
0120 
0121 ipmi_poweroff - Some systems support the ability to be turned off via
0122 IPMI commands.
0123 
0124 bt-bmc - This is not part of the main driver, but instead a driver for
0125 accessing a BMC-side interface of a BT interface.  It is used on BMCs
0126 running Linux to provide an interface to the host.
0127 
0128 These are all individually selectable via configuration options.
0129 
0130 Much documentation for the interface is in the include files.  The
0131 IPMI include files are:
0132 
0133 linux/ipmi.h - Contains the user interface and IOCTL interface for IPMI.
0134 
0135 linux/ipmi_smi.h - Contains the interface for system management interfaces
0136 (things that interface to IPMI controllers) to use.
0137 
0138 linux/ipmi_msgdefs.h - General definitions for base IPMI messaging.
0139 
0140 
0141 Addressing
0142 ----------
0143 
0144 The IPMI addressing works much like IP addresses, you have an overlay
0145 to handle the different address types.  The overlay is::
0146 
0147   struct ipmi_addr
0148   {
0149         int   addr_type;
0150         short channel;
0151         char  data[IPMI_MAX_ADDR_SIZE];
0152   };
0153 
0154 The addr_type determines what the address really is.  The driver
0155 currently understands two different types of addresses.
0156 
0157 "System Interface" addresses are defined as::
0158 
0159   struct ipmi_system_interface_addr
0160   {
0161         int   addr_type;
0162         short channel;
0163   };
0164 
0165 and the type is IPMI_SYSTEM_INTERFACE_ADDR_TYPE.  This is used for talking
0166 straight to the BMC on the current card.  The channel must be
0167 IPMI_BMC_CHANNEL.
0168 
0169 Messages that are destined to go out on the IPMB bus going through the
0170 BMC use the IPMI_IPMB_ADDR_TYPE address type.  The format is::
0171 
0172   struct ipmi_ipmb_addr
0173   {
0174         int           addr_type;
0175         short         channel;
0176         unsigned char slave_addr;
0177         unsigned char lun;
0178   };
0179 
0180 The "channel" here is generally zero, but some devices support more
0181 than one channel, it corresponds to the channel as defined in the IPMI
0182 spec.
0183 
0184 There is also an IPMB direct address for a situation where the sender
0185 is directly on an IPMB bus and doesn't have to go through the BMC.
0186 You can send messages to a specific management controller (MC) on the
0187 IPMB using the IPMI_IPMB_DIRECT_ADDR_TYPE with the following format::
0188 
0189   struct ipmi_ipmb_direct_addr
0190   {
0191         int           addr_type;
0192         short         channel;
0193         unsigned char slave_addr;
0194         unsigned char rq_lun;
0195         unsigned char rs_lun;
0196   };
0197 
0198 The channel is always zero.  You can also receive commands from other
0199 MCs that you have registered to handle and respond to them, so you can
0200 use this to implement a management controller on a bus..
0201 
0202 Messages
0203 --------
0204 
0205 Messages are defined as::
0206 
0207   struct ipmi_msg
0208   {
0209         unsigned char netfn;
0210         unsigned char lun;
0211         unsigned char cmd;
0212         unsigned char *data;
0213         int           data_len;
0214   };
0215 
0216 The driver takes care of adding/stripping the header information.  The
0217 data portion is just the data to be send (do NOT put addressing info
0218 here) or the response.  Note that the completion code of a response is
0219 the first item in "data", it is not stripped out because that is how
0220 all the messages are defined in the spec (and thus makes counting the
0221 offsets a little easier :-).
0222 
0223 When using the IOCTL interface from userland, you must provide a block
0224 of data for "data", fill it, and set data_len to the length of the
0225 block of data, even when receiving messages.  Otherwise the driver
0226 will have no place to put the message.
0227 
0228 Messages coming up from the message handler in kernelland will come in
0229 as::
0230 
0231   struct ipmi_recv_msg
0232   {
0233         struct list_head link;
0234 
0235         /* The type of message as defined in the "Receive Types"
0236            defines above. */
0237         int         recv_type;
0238 
0239         ipmi_user_t      *user;
0240         struct ipmi_addr addr;
0241         long             msgid;
0242         struct ipmi_msg  msg;
0243 
0244         /* Call this when done with the message.  It will presumably free
0245            the message and do any other necessary cleanup. */
0246         void (*done)(struct ipmi_recv_msg *msg);
0247 
0248         /* Place-holder for the data, don't make any assumptions about
0249            the size or existence of this, since it may change. */
0250         unsigned char   msg_data[IPMI_MAX_MSG_LENGTH];
0251   };
0252 
0253 You should look at the receive type and handle the message
0254 appropriately.
0255 
0256 
0257 The Upper Layer Interface (Message Handler)
0258 -------------------------------------------
0259 
0260 The upper layer of the interface provides the users with a consistent
0261 view of the IPMI interfaces.  It allows multiple SMI interfaces to be
0262 addressed (because some boards actually have multiple BMCs on them)
0263 and the user should not have to care what type of SMI is below them.
0264 
0265 
0266 Watching For Interfaces
0267 ^^^^^^^^^^^^^^^^^^^^^^^
0268 
0269 When your code comes up, the IPMI driver may or may not have detected
0270 if IPMI devices exist.  So you might have to defer your setup until
0271 the device is detected, or you might be able to do it immediately.
0272 To handle this, and to allow for discovery, you register an SMI
0273 watcher with ipmi_smi_watcher_register() to iterate over interfaces
0274 and tell you when they come and go.
0275 
0276 
0277 Creating the User
0278 ^^^^^^^^^^^^^^^^^
0279 
0280 To use the message handler, you must first create a user using
0281 ipmi_create_user.  The interface number specifies which SMI you want
0282 to connect to, and you must supply callback functions to be called
0283 when data comes in.  The callback function can run at interrupt level,
0284 so be careful using the callbacks.  This also allows to you pass in a
0285 piece of data, the handler_data, that will be passed back to you on
0286 all calls.
0287 
0288 Once you are done, call ipmi_destroy_user() to get rid of the user.
0289 
0290 From userland, opening the device automatically creates a user, and
0291 closing the device automatically destroys the user.
0292 
0293 
0294 Messaging
0295 ^^^^^^^^^
0296 
0297 To send a message from kernel-land, the ipmi_request_settime() call does
0298 pretty much all message handling.  Most of the parameter are
0299 self-explanatory.  However, it takes a "msgid" parameter.  This is NOT
0300 the sequence number of messages.  It is simply a long value that is
0301 passed back when the response for the message is returned.  You may
0302 use it for anything you like.
0303 
0304 Responses come back in the function pointed to by the ipmi_recv_hndl
0305 field of the "handler" that you passed in to ipmi_create_user().
0306 Remember again, these may be running at interrupt level.  Remember to
0307 look at the receive type, too.
0308 
0309 From userland, you fill out an ipmi_req_t structure and use the
0310 IPMICTL_SEND_COMMAND ioctl.  For incoming stuff, you can use select()
0311 or poll() to wait for messages to come in.  However, you cannot use
0312 read() to get them, you must call the IPMICTL_RECEIVE_MSG with the
0313 ipmi_recv_t structure to actually get the message.  Remember that you
0314 must supply a pointer to a block of data in the msg.data field, and
0315 you must fill in the msg.data_len field with the size of the data.
0316 This gives the receiver a place to actually put the message.
0317 
0318 If the message cannot fit into the data you provide, you will get an
0319 EMSGSIZE error and the driver will leave the data in the receive
0320 queue.  If you want to get it and have it truncate the message, us
0321 the IPMICTL_RECEIVE_MSG_TRUNC ioctl.
0322 
0323 When you send a command (which is defined by the lowest-order bit of
0324 the netfn per the IPMI spec) on the IPMB bus, the driver will
0325 automatically assign the sequence number to the command and save the
0326 command.  If the response is not receive in the IPMI-specified 5
0327 seconds, it will generate a response automatically saying the command
0328 timed out.  If an unsolicited response comes in (if it was after 5
0329 seconds, for instance), that response will be ignored.
0330 
0331 In kernelland, after you receive a message and are done with it, you
0332 MUST call ipmi_free_recv_msg() on it, or you will leak messages.  Note
0333 that you should NEVER mess with the "done" field of a message, that is
0334 required to properly clean up the message.
0335 
0336 Note that when sending, there is an ipmi_request_supply_msgs() call
0337 that lets you supply the smi and receive message.  This is useful for
0338 pieces of code that need to work even if the system is out of buffers
0339 (the watchdog timer uses this, for instance).  You supply your own
0340 buffer and own free routines.  This is not recommended for normal use,
0341 though, since it is tricky to manage your own buffers.
0342 
0343 
0344 Events and Incoming Commands
0345 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
0346 
0347 The driver takes care of polling for IPMI events and receiving
0348 commands (commands are messages that are not responses, they are
0349 commands that other things on the IPMB bus have sent you).  To receive
0350 these, you must register for them, they will not automatically be sent
0351 to you.
0352 
0353 To receive events, you must call ipmi_set_gets_events() and set the
0354 "val" to non-zero.  Any events that have been received by the driver
0355 since startup will immediately be delivered to the first user that
0356 registers for events.  After that, if multiple users are registered
0357 for events, they will all receive all events that come in.
0358 
0359 For receiving commands, you have to individually register commands you
0360 want to receive.  Call ipmi_register_for_cmd() and supply the netfn
0361 and command name for each command you want to receive.  You also
0362 specify a bitmask of the channels you want to receive the command from
0363 (or use IPMI_CHAN_ALL for all channels if you don't care).  Only one
0364 user may be registered for each netfn/cmd/channel, but different users
0365 may register for different commands, or the same command if the
0366 channel bitmasks do not overlap.
0367 
0368 To respond to a received command, set the response bit in the returned
0369 netfn, use the address from the received message, and use the same
0370 msgid that you got in the receive message.
0371 
0372 From userland, equivalent IOCTLs are provided to do these functions.
0373 
0374 
0375 The Lower Layer (SMI) Interface
0376 -------------------------------
0377 
0378 As mentioned before, multiple SMI interfaces may be registered to the
0379 message handler, each of these is assigned an interface number when
0380 they register with the message handler.  They are generally assigned
0381 in the order they register, although if an SMI unregisters and then
0382 another one registers, all bets are off.
0383 
0384 The ipmi_smi.h defines the interface for management interfaces, see
0385 that for more details.
0386 
0387 
0388 The SI Driver
0389 -------------
0390 
0391 The SI driver allows KCS, BT, and SMIC interfaces to be configured
0392 in the system.  It discovers interfaces through a host of different
0393 methods, depending on the system.
0394 
0395 You can specify up to four interfaces on the module load line and
0396 control some module parameters::
0397 
0398   modprobe ipmi_si.o type=<type1>,<type2>....
0399        ports=<port1>,<port2>... addrs=<addr1>,<addr2>...
0400        irqs=<irq1>,<irq2>...
0401        regspacings=<sp1>,<sp2>,... regsizes=<size1>,<size2>,...
0402        regshifts=<shift1>,<shift2>,...
0403        slave_addrs=<addr1>,<addr2>,...
0404        force_kipmid=<enable1>,<enable2>,...
0405        kipmid_max_busy_us=<ustime1>,<ustime2>,...
0406        unload_when_empty=[0|1]
0407        trydmi=[0|1] tryacpi=[0|1]
0408        tryplatform=[0|1] trypci=[0|1]
0409 
0410 Each of these except try... items is a list, the first item for the
0411 first interface, second item for the second interface, etc.
0412 
0413 The si_type may be either "kcs", "smic", or "bt".  If you leave it blank, it
0414 defaults to "kcs".
0415 
0416 If you specify addrs as non-zero for an interface, the driver will
0417 use the memory address given as the address of the device.  This
0418 overrides si_ports.
0419 
0420 If you specify ports as non-zero for an interface, the driver will
0421 use the I/O port given as the device address.
0422 
0423 If you specify irqs as non-zero for an interface, the driver will
0424 attempt to use the given interrupt for the device.
0425 
0426 The other try... items disable discovery by their corresponding
0427 names.  These are all enabled by default, set them to zero to disable
0428 them.  The tryplatform disables openfirmware.
0429 
0430 The next three parameters have to do with register layout.  The
0431 registers used by the interfaces may not appear at successive
0432 locations and they may not be in 8-bit registers.  These parameters
0433 allow the layout of the data in the registers to be more precisely
0434 specified.
0435 
0436 The regspacings parameter give the number of bytes between successive
0437 register start addresses.  For instance, if the regspacing is set to 4
0438 and the start address is 0xca2, then the address for the second
0439 register would be 0xca6.  This defaults to 1.
0440 
0441 The regsizes parameter gives the size of a register, in bytes.  The
0442 data used by IPMI is 8-bits wide, but it may be inside a larger
0443 register.  This parameter allows the read and write type to specified.
0444 It may be 1, 2, 4, or 8.  The default is 1.
0445 
0446 Since the register size may be larger than 32 bits, the IPMI data may not
0447 be in the lower 8 bits.  The regshifts parameter give the amount to shift
0448 the data to get to the actual IPMI data.
0449 
0450 The slave_addrs specifies the IPMI address of the local BMC.  This is
0451 usually 0x20 and the driver defaults to that, but in case it's not, it
0452 can be specified when the driver starts up.
0453 
0454 The force_ipmid parameter forcefully enables (if set to 1) or disables
0455 (if set to 0) the kernel IPMI daemon.  Normally this is auto-detected
0456 by the driver, but systems with broken interrupts might need an enable,
0457 or users that don't want the daemon (don't need the performance, don't
0458 want the CPU hit) can disable it.
0459 
0460 If unload_when_empty is set to 1, the driver will be unloaded if it
0461 doesn't find any interfaces or all the interfaces fail to work.  The
0462 default is one.  Setting to 0 is useful with the hotmod, but is
0463 obviously only useful for modules.
0464 
0465 When compiled into the kernel, the parameters can be specified on the
0466 kernel command line as::
0467 
0468   ipmi_si.type=<type1>,<type2>...
0469        ipmi_si.ports=<port1>,<port2>... ipmi_si.addrs=<addr1>,<addr2>...
0470        ipmi_si.irqs=<irq1>,<irq2>...
0471        ipmi_si.regspacings=<sp1>,<sp2>,...
0472        ipmi_si.regsizes=<size1>,<size2>,...
0473        ipmi_si.regshifts=<shift1>,<shift2>,...
0474        ipmi_si.slave_addrs=<addr1>,<addr2>,...
0475        ipmi_si.force_kipmid=<enable1>,<enable2>,...
0476        ipmi_si.kipmid_max_busy_us=<ustime1>,<ustime2>,...
0477 
0478 It works the same as the module parameters of the same names.
0479 
0480 If your IPMI interface does not support interrupts and is a KCS or
0481 SMIC interface, the IPMI driver will start a kernel thread for the
0482 interface to help speed things up.  This is a low-priority kernel
0483 thread that constantly polls the IPMI driver while an IPMI operation
0484 is in progress.  The force_kipmid module parameter will all the user to
0485 force this thread on or off.  If you force it off and don't have
0486 interrupts, the driver will run VERY slowly.  Don't blame me,
0487 these interfaces suck.
0488 
0489 Unfortunately, this thread can use a lot of CPU depending on the
0490 interface's performance.  This can waste a lot of CPU and cause
0491 various issues with detecting idle CPU and using extra power.  To
0492 avoid this, the kipmid_max_busy_us sets the maximum amount of time, in
0493 microseconds, that kipmid will spin before sleeping for a tick.  This
0494 value sets a balance between performance and CPU waste and needs to be
0495 tuned to your needs.  Maybe, someday, auto-tuning will be added, but
0496 that's not a simple thing and even the auto-tuning would need to be
0497 tuned to the user's desired performance.
0498 
0499 The driver supports a hot add and remove of interfaces.  This way,
0500 interfaces can be added or removed after the kernel is up and running.
0501 This is done using /sys/modules/ipmi_si/parameters/hotmod, which is a
0502 write-only parameter.  You write a string to this interface.  The string
0503 has the format::
0504 
0505    <op1>[:op2[:op3...]]
0506 
0507 The "op"s are::
0508 
0509    add|remove,kcs|bt|smic,mem|i/o,<address>[,<opt1>[,<opt2>[,...]]]
0510 
0511 You can specify more than one interface on the line.  The "opt"s are::
0512 
0513    rsp=<regspacing>
0514    rsi=<regsize>
0515    rsh=<regshift>
0516    irq=<irq>
0517    ipmb=<ipmb slave addr>
0518 
0519 and these have the same meanings as discussed above.  Note that you
0520 can also use this on the kernel command line for a more compact format
0521 for specifying an interface.  Note that when removing an interface,
0522 only the first three parameters (si type, address type, and address)
0523 are used for the comparison.  Any options are ignored for removing.
0524 
0525 The SMBus Driver (SSIF)
0526 -----------------------
0527 
0528 The SMBus driver allows up to 4 SMBus devices to be configured in the
0529 system.  By default, the driver will only register with something it
0530 finds in DMI or ACPI tables.  You can change this
0531 at module load time (for a module) with::
0532 
0533   modprobe ipmi_ssif.o
0534         addr=<i2caddr1>[,<i2caddr2>[,...]]
0535         adapter=<adapter1>[,<adapter2>[...]]
0536         dbg=<flags1>,<flags2>...
0537         slave_addrs=<addr1>,<addr2>,...
0538         tryacpi=[0|1] trydmi=[0|1]
0539         [dbg_probe=1]
0540         alerts_broken
0541 
0542 The addresses are normal I2C addresses.  The adapter is the string
0543 name of the adapter, as shown in /sys/class/i2c-adapter/i2c-<n>/name.
0544 It is *NOT* i2c-<n> itself.  Also, the comparison is done ignoring
0545 spaces, so if the name is "This is an I2C chip" you can say
0546 adapter_name=ThisisanI2cchip.  This is because it's hard to pass in
0547 spaces in kernel parameters.
0548 
0549 The debug flags are bit flags for each BMC found, they are:
0550 IPMI messages: 1, driver state: 2, timing: 4, I2C probe: 8
0551 
0552 The tryxxx parameters can be used to disable detecting interfaces
0553 from various sources.
0554 
0555 Setting dbg_probe to 1 will enable debugging of the probing and
0556 detection process for BMCs on the SMBusses.
0557 
0558 The slave_addrs specifies the IPMI address of the local BMC.  This is
0559 usually 0x20 and the driver defaults to that, but in case it's not, it
0560 can be specified when the driver starts up.
0561 
0562 alerts_broken does not enable SMBus alert for SSIF. Otherwise SMBus
0563 alert will be enabled on supported hardware.
0564 
0565 Discovering the IPMI compliant BMC on the SMBus can cause devices on
0566 the I2C bus to fail. The SMBus driver writes a "Get Device ID" IPMI
0567 message as a block write to the I2C bus and waits for a response.
0568 This action can be detrimental to some I2C devices. It is highly
0569 recommended that the known I2C address be given to the SMBus driver in
0570 the smb_addr parameter unless you have DMI or ACPI data to tell the
0571 driver what to use.
0572 
0573 When compiled into the kernel, the addresses can be specified on the
0574 kernel command line as::
0575 
0576   ipmb_ssif.addr=<i2caddr1>[,<i2caddr2>[...]]
0577         ipmi_ssif.adapter=<adapter1>[,<adapter2>[...]]
0578         ipmi_ssif.dbg=<flags1>[,<flags2>[...]]
0579         ipmi_ssif.dbg_probe=1
0580         ipmi_ssif.slave_addrs=<addr1>[,<addr2>[...]]
0581         ipmi_ssif.tryacpi=[0|1] ipmi_ssif.trydmi=[0|1]
0582 
0583 These are the same options as on the module command line.
0584 
0585 The I2C driver does not support non-blocking access or polling, so
0586 this driver cannod to IPMI panic events, extend the watchdog at panic
0587 time, or other panic-related IPMI functions without special kernel
0588 patches and driver modifications.  You can get those at the openipmi
0589 web page.
0590 
0591 The driver supports a hot add and remove of interfaces through the I2C
0592 sysfs interface.
0593 
0594 The IPMI IPMB Driver
0595 --------------------
0596 
0597 This driver is for supporting a system that sits on an IPMB bus; it
0598 allows the interface to look like a normal IPMI interface.  Sending
0599 system interface addressed messages to it will cause the message to go
0600 to the registered BMC on the system (default at IPMI address 0x20).
0601 
0602 It also allows you to directly address other MCs on the bus using the
0603 ipmb direct addressing.  You can receive commands from other MCs on
0604 the bus and they will be handled through the normal received command
0605 mechanism described above.
0606 
0607 Parameters are::
0608 
0609   ipmi_ipmb.bmcaddr=<address to use for system interface addresses messages>
0610         ipmi_ipmb.retry_time_ms=<Time between retries on IPMB>
0611         ipmi_ipmb.max_retries=<Number of times to retry a message>
0612 
0613 Loading the module will not result in the driver automatcially
0614 starting unless there is device tree information setting it up.  If
0615 you want to instantiate one of these by hand, do::
0616 
0617   echo ipmi-ipmb <addr> > /sys/class/i2c-dev/i2c-<n>/device/new_device
0618 
0619 Note that the address you give here is the I2C address, not the IPMI
0620 address.  So if you want your MC address to be 0x60, you put 0x30
0621 here.  See the I2C driver info for more details.
0622 
0623 Command bridging to other IPMB busses through this interface does not
0624 work.  The receive message queue is not implemented, by design.  There
0625 is only one receive message queue on a BMC, and that is meant for the
0626 host drivers, not something on the IPMB bus.
0627 
0628 A BMC may have multiple IPMB busses, which bus your device sits on
0629 depends on how the system is wired.  You can fetch the channels with
0630 "ipmitool channel info <n>" where <n> is the channel, with the
0631 channels being 0-7 and try the IPMB channels.
0632 
0633 Other Pieces
0634 ------------
0635 
0636 Get the detailed info related with the IPMI device
0637 --------------------------------------------------
0638 
0639 Some users need more detailed information about a device, like where
0640 the address came from or the raw base device for the IPMI interface.
0641 You can use the IPMI smi_watcher to catch the IPMI interfaces as they
0642 come or go, and to grab the information, you can use the function
0643 ipmi_get_smi_info(), which returns the following structure::
0644 
0645   struct ipmi_smi_info {
0646         enum ipmi_addr_src addr_src;
0647         struct device *dev;
0648         union {
0649                 struct {
0650                         void *acpi_handle;
0651                 } acpi_info;
0652         } addr_info;
0653   };
0654 
0655 Currently special info for only for SI_ACPI address sources is
0656 returned.  Others may be added as necessary.
0657 
0658 Note that the dev pointer is included in the above structure, and
0659 assuming ipmi_smi_get_info returns success, you must call put_device
0660 on the dev pointer.
0661 
0662 
0663 Watchdog
0664 --------
0665 
0666 A watchdog timer is provided that implements the Linux-standard
0667 watchdog timer interface.  It has three module parameters that can be
0668 used to control it::
0669 
0670   modprobe ipmi_watchdog timeout=<t> pretimeout=<t> action=<action type>
0671       preaction=<preaction type> preop=<preop type> start_now=x
0672       nowayout=x ifnum_to_use=n panic_wdt_timeout=<t>
0673 
0674 ifnum_to_use specifies which interface the watchdog timer should use.
0675 The default is -1, which means to pick the first one registered.
0676 
0677 The timeout is the number of seconds to the action, and the pretimeout
0678 is the amount of seconds before the reset that the pre-timeout panic will
0679 occur (if pretimeout is zero, then pretimeout will not be enabled).  Note
0680 that the pretimeout is the time before the final timeout.  So if the
0681 timeout is 50 seconds and the pretimeout is 10 seconds, then the pretimeout
0682 will occur in 40 second (10 seconds before the timeout). The panic_wdt_timeout
0683 is the value of timeout which is set on kernel panic, in order to let actions
0684 such as kdump to occur during panic.
0685 
0686 The action may be "reset", "power_cycle", or "power_off", and
0687 specifies what to do when the timer times out, and defaults to
0688 "reset".
0689 
0690 The preaction may be "pre_smi" for an indication through the SMI
0691 interface, "pre_int" for an indication through the SMI with an
0692 interrupts, and "pre_nmi" for a NMI on a preaction.  This is how
0693 the driver is informed of the pretimeout.
0694 
0695 The preop may be set to "preop_none" for no operation on a pretimeout,
0696 "preop_panic" to set the preoperation to panic, or "preop_give_data"
0697 to provide data to read from the watchdog device when the pretimeout
0698 occurs.  A "pre_nmi" setting CANNOT be used with "preop_give_data"
0699 because you can't do data operations from an NMI.
0700 
0701 When preop is set to "preop_give_data", one byte comes ready to read
0702 on the device when the pretimeout occurs.  Select and fasync work on
0703 the device, as well.
0704 
0705 If start_now is set to 1, the watchdog timer will start running as
0706 soon as the driver is loaded.
0707 
0708 If nowayout is set to 1, the watchdog timer will not stop when the
0709 watchdog device is closed.  The default value of nowayout is true
0710 if the CONFIG_WATCHDOG_NOWAYOUT option is enabled, or false if not.
0711 
0712 When compiled into the kernel, the kernel command line is available
0713 for configuring the watchdog::
0714 
0715   ipmi_watchdog.timeout=<t> ipmi_watchdog.pretimeout=<t>
0716         ipmi_watchdog.action=<action type>
0717         ipmi_watchdog.preaction=<preaction type>
0718         ipmi_watchdog.preop=<preop type>
0719         ipmi_watchdog.start_now=x
0720         ipmi_watchdog.nowayout=x
0721         ipmi_watchdog.panic_wdt_timeout=<t>
0722 
0723 The options are the same as the module parameter options.
0724 
0725 The watchdog will panic and start a 120 second reset timeout if it
0726 gets a pre-action.  During a panic or a reboot, the watchdog will
0727 start a 120 timer if it is running to make sure the reboot occurs.
0728 
0729 Note that if you use the NMI preaction for the watchdog, you MUST NOT
0730 use the nmi watchdog.  There is no reasonable way to tell if an NMI
0731 comes from the IPMI controller, so it must assume that if it gets an
0732 otherwise unhandled NMI, it must be from IPMI and it will panic
0733 immediately.
0734 
0735 Once you open the watchdog timer, you must write a 'V' character to the
0736 device to close it, or the timer will not stop.  This is a new semantic
0737 for the driver, but makes it consistent with the rest of the watchdog
0738 drivers in Linux.
0739 
0740 
0741 Panic Timeouts
0742 --------------
0743 
0744 The OpenIPMI driver supports the ability to put semi-custom and custom
0745 events in the system event log if a panic occurs.  if you enable the
0746 'Generate a panic event to all BMCs on a panic' option, you will get
0747 one event on a panic in a standard IPMI event format.  If you enable
0748 the 'Generate OEM events containing the panic string' option, you will
0749 also get a bunch of OEM events holding the panic string.
0750 
0751 
0752 The field settings of the events are:
0753 
0754 * Generator ID: 0x21 (kernel)
0755 * EvM Rev: 0x03 (this event is formatting in IPMI 1.0 format)
0756 * Sensor Type: 0x20 (OS critical stop sensor)
0757 * Sensor #: The first byte of the panic string (0 if no panic string)
0758 * Event Dir | Event Type: 0x6f (Assertion, sensor-specific event info)
0759 * Event Data 1: 0xa1 (Runtime stop in OEM bytes 2 and 3)
0760 * Event data 2: second byte of panic string
0761 * Event data 3: third byte of panic string
0762 
0763 See the IPMI spec for the details of the event layout.  This event is
0764 always sent to the local management controller.  It will handle routing
0765 the message to the right place
0766 
0767 Other OEM events have the following format:
0768 
0769 * Record ID (bytes 0-1): Set by the SEL.
0770 * Record type (byte 2): 0xf0 (OEM non-timestamped)
0771 * byte 3: The slave address of the card saving the panic
0772 * byte 4: A sequence number (starting at zero)
0773   The rest of the bytes (11 bytes) are the panic string.  If the panic string
0774   is longer than 11 bytes, multiple messages will be sent with increasing
0775   sequence numbers.
0776 
0777 Because you cannot send OEM events using the standard interface, this
0778 function will attempt to find an SEL and add the events there.  It
0779 will first query the capabilities of the local management controller.
0780 If it has an SEL, then they will be stored in the SEL of the local
0781 management controller.  If not, and the local management controller is
0782 an event generator, the event receiver from the local management
0783 controller will be queried and the events sent to the SEL on that
0784 device.  Otherwise, the events go nowhere since there is nowhere to
0785 send them.
0786 
0787 
0788 Poweroff
0789 --------
0790 
0791 If the poweroff capability is selected, the IPMI driver will install
0792 a shutdown function into the standard poweroff function pointer.  This
0793 is in the ipmi_poweroff module.  When the system requests a powerdown,
0794 it will send the proper IPMI commands to do this.  This is supported on
0795 several platforms.
0796 
0797 There is a module parameter named "poweroff_powercycle" that may
0798 either be zero (do a power down) or non-zero (do a power cycle, power
0799 the system off, then power it on in a few seconds).  Setting
0800 ipmi_poweroff.poweroff_control=x will do the same thing on the kernel
0801 command line.  The parameter is also available via the proc filesystem
0802 in /proc/sys/dev/ipmi/poweroff_powercycle.  Note that if the system
0803 does not support power cycling, it will always do the power off.
0804 
0805 The "ifnum_to_use" parameter specifies which interface the poweroff
0806 code should use.  The default is -1, which means to pick the first one
0807 registered.
0808 
0809 Note that if you have ACPI enabled, the system will prefer using ACPI to
0810 power off.