xref: /titanic_44/usr/src/man/man1m/cfgadm_sbd.1m (revision d4f95bf4d6ec7ef0f01e5ddf22813ac641edf019)
te
Copyright 2003 (c), Sun Microsystems, Inc. All Rights Reserved
The contents of this file are subject to the terms of the Common Development and Distribution License (the "License"). You may not use this file except in compliance with the License.
You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE or http://www.opensolaris.org/os/licensing. See the License for the specific language governing permissions and limitations under the License.
When distributing Covered Code, include this CDDL HEADER in each file and include the License file at usr/src/OPENSOLARIS.LICENSE. If applicable, add the following below this CDDL HEADER, with the fields enclosed by brackets "[]" replaced with your own identifying information: Portions Copyright [yyyy] [name of copyright owner]
CFGADM_SBD 1M "Oct 13, 2003"
NAME
cfgadm_sbd - cfgadm commands for system board administration
SYNOPSIS

cfgadm  -l [-a] [-o parsable]  ap_id...

cfgadm  -c  function [-f] [-y | -n]
 [-o unassign | nopoweroff] [-v]  ap_id...

cfgadm  -t [-v]  ap_id...

cfgadm  -x  [-f] [-v] function  ap_id...
DESCRIPTION

The cfgadm_sbd plugin provides dynamic reconfiguration functionality for connecting, configuring, unconfiguring, and disconnecting class sbd system boards. It also enables you to connect or disconnect a system board from a running system without having to reboot the system.

The cfgadm command resides in /usr/sbin. See cfgadm(1M). The cfgadm_sbd plugin resides /usr/platform/sun4u/lib/cfgadm.

Each board slot appears as a single attachment point in the device tree. Each component appears as a dynamic attachment point. You can view the type, state, and condition of each component, and the states and condition of each board slot by using the -a option.

The cfgadm options perform differently depending on the platform. Additionally, the form of the attachment points is different depending on the platform. See the Platform Notes section for more information.

"Component Conditions"

The following are the names and descriptions of the component conditions: failed

The component failed testing.

ok

The component is operational.

unknown

The component has not been tested.

"Component States"

The following is the name and description of the receptacle state for components: connected

The component is connected to the board slot.

The following are the names and descriptions of the occupant states for components: configured

The component is available for use by the Solaris operating environment.

unconfigured

The component is not available for use by the Solaris operating environment.

"Board Conditions"

The following are the names and descriptions of the board conditions. failed

The board failed testing.

ok

The board is operational.

unknown

The board has not been tested.

unusable

The board slot is unusable.

"Board States"

Inserting a board changes the receptacle state from empty to disconnected. Removing a board changes the receptacle state from disconnected to empty.

Caution: Removing a board that is in the connected state or that is powered on and in the disconnected state crashes the operating system and can result in permanent damage to the system.

The following are the names and descriptions of the receptacle states for boards: connected

The board is powered on and connected to the system bus. You can view the components on a board only after it is in the connected state.

disconnected

The board is disconnected from the system bus. A board can be in the disconnected state without being powered off. However, a board must be powered off and in the disconnected state before you remove it from the slot.

empty

A board is not present.

The occupant state of a disconnected board is always unconfigured. The following table contains the names and descriptions of the occupant states for boards: configured

At least one component on the board is configured.

unconfigured

All of the components on the board are unconfigured.

"Dynamic System Domains"

Platforms based on dynamic system domains (DSDs, referred to as domains in this document) divide the slots in the chassis into electrically isolated hardware partitions (that is, DSDs). Platforms that are not based on DSDs assign all slots to the system permanently.

A slot can be empty or populated, and it can be assigned or available to any number of domains. The number of slots available to a given domain is controlled by an available component list (ACL) that is maintained on the system controller. The ACL is not the access control list provided by the Solaris operating environment.

A slot is visible to a domain only if the slot is in the domain's ACL and if it is not assigned to another domain. An unassigned slot is visible to all domains that have the slot in their ACL. After a slot has been assigned to a domain, the slot is no longer visible to any other domain.

A slot that is visible to a domain, but not assigned, must first be assigned to the domain before any other state changing commands are applied. The assign can be done explicitly using -x assign or implicitly as part of a connect. A slot must be unassigned from a domain before it can be used by another domain. The unassign is always explicit, either directly using -x unassign or as an option to disconnect using -o unassign.

"State Change Functions"

Functions that change the state of a board slot or a component on the board can be issued concurrently against any attachment point. Only one state changing operation is permitted at a given time. A Y in the Busy field in the state changing information indicates an operation is in progress.

The following list contains the functions that change the state:

configure

unconfigure

connect

disconnect

"Availability Change Functions"

Commands that change the availability of a board can be issued concurrently against any attachment point. Only one availability change operation is permitted at a given time. These functions also change the information string in the cfgadm -l output. A Y in the Busy field indicates that an operation is in progress.

The following list contains the functions that change the availability:

assign

unassign

"Condition Change Functions"

Functions that change the condition of a board slot or a component on the board can be issued concurrently against any attachment point. Only one condition change operation is permitted at a given time. These functions also change the information string in the cfgadm -l output. A Y in the Busy field indicates an operation is in progress.

The following list contains the functions that change the condition:

poweron

poweroff

test

"Unconfigure Process"

This section contains a description of the unconfigure process, and illustrates the states of source and target boards at different stages during the process of moving permanent memory.

In the following code examples, the permanent memory on board 0 must be moved to another board in the domain. Thus, board 0 is the source, and board 1 is the target.

A status change operation cannot be initiated on a board while it is marked as busy. For brevity, the CPU information has been removed from the code examples.

The process is started with the following command:

# cfgadm -c unconfigure -y SB0::memory &

First, the memory on board 1 in the same address range as the permanent memory on board 0 must be deleted. During this phase, the source board, the target board, and the memory attachment points are marked as busy. You can display the status with the following command:

# cfgadm -a -s cols=ap_id:type:r_state:o_state:busy SB0 SB1

Ap_Id Type Receptacle Occupant Busy
SB0 CPU connected configured y
SB0::memory memory connected configured y
SB1 CPU connected configured y
SB1::memory memory connected configured y

After the memory has been deleted on board 1, it is marked as unconfigured. The memory on board 0 remains configured, but it is still marked as busy, as in the following example.

Ap_Id Type Receptacle Occupant Busy
SB0 CPU connected configured y
SB0::memory memory connected configured y
SB1 CPU connected configured y
SB1::memory memory connected unconfigured n

The memory from board 0 is then copied to board 1. After it has been copied, the occupant state for the memory is switched. The memory on board 0 becomes unconfigured, and the memory on board 1 becomes configured. At this point in the process, only board 0 remains busy, as in the following example.

Ap_Id Type Receptacle Occupant Busy
SB0 CPU connected configured y
SB0::memory memory connected unconfigured n
SB1 CPU connected configured n
SB1::memory memory connected configured n

After the entire process has been completed, the memory on board 0 remains unconfigured, and the attachment points are not busy, as in the following example.

Ap_Id Type Receptacle Occupant Busy
SB0 CPU connected configured n
SB0::memory memory connected unconfigured n
SB1 CPU connected configured n
SB1::memory memory connected configured n

The permanent memory has been moved, and the memory on board 0 has been unconfigured. At this point, you can initiate a new state changing operation on either board.

"Platform-Specific Options"

You can specify platform-specific options that follow the options interpreted by the system board plugin. All platform-specific options must be preceded by the platform keyword. The following example contains the general format of a command with platform-specific options:

command -o sbd_options,platform=platform_options

OPTIONS

This man page does not include the -v, -a, -s, or -h options for the cfgadm command. See cfgadm(1M) for descriptions of those options. The following options are supported by the cfgadm_sbd plugin: -c function

Performs a state change function. You can use the following functions: unconfigure

Changes the occupant state to unconfigured. This function applies to system board slots and to all of the components on the system board. The unconfigure function removes the CPUs from the CPU list and deletes the physical memory from the system memory pool. If any device is still in use, the cfgadm command fails and reports the failure to the user. You can retry the command as soon as the device is no longer busy. If a CPU is in use, you must ensure that it is off line before you proceed. See pbind(1M), psradm(1M) and psrinfo(1M). The unconfigure function moves the physical memory to another system board before it deletes the memory from the board you want to unconfigure. Depending of the type of memory being moved, the command fails if it cannot find enough memory on another board or if it cannot find an appropriate physical memory range. For permanent memory, the operating system must be suspended (that is, quiesced) while the memory is moved and the memory controllers are reprogrammed. If the operating system must be suspended, you will be prompted to proceed with the operation. You can use the -y or -n options to always answer yes or no respectively. Moving memory can take several minutes to complete, depending on the amount of memory and the system load. You can monitor the progress of the operation by issuing a status command against the memory attachment point. You can also interrupt the memory operation by stopping the cfgadm command. The deleted memory is returned to the system memory pool.

disconnect

Changes the receptacle state to disconnected. This function applies only to system board slots. If the occupant state is configured, the disconnect function attempts to unconfigure the occupant. It then powers off the system board. At this point, the board can be removed from the slot. This function leaves the board in the assigned state on platforms that support dynamic system domains. If you specify -o nopoweroff, the disconnect function leaves the board powered on. If you specify -o unassign, the disconnect function unassigns the board from the domain. If you unassign a board from a domain, you can assign it to another domain. However, if it is assigned to another domain, it is not available to the domain from which is was unassigned.

configure

Changes the occupant state to configured. This function applies to system board slots and to any components on the system board. If the receptacle state is disconnected, the configure function attempts to connect the receptacle. It then walks the tree of devices that is created by the connect function, and attaches the devices if necessary. Running this function configures all of the components on the board, except those that have already been configured. For CPUs, the configure function adds the CPUs to the CPU list. For memory, the configure function ensures that the memory is initialized then adds the memory to the system memory pool. The CPUs and the memory are ready for use after the configure function has been completed successfully. For I/O devices, you must use the mount and the ifconfig commands before the devices can be used. See ifconfig(1M) and mount(1M).

connect

Changes the receptacle state to connected. This function applies only to system board slots. If the board slot is not assigned to the domain, the connect function attempts to assign the slot to the domain. Next, it powers on and tests the board, then it connects the board electronically to the system bus and probes the components. After the connect function is completed successfully, you can use the -a option to view the status of the components on the board. The connect function leaves all of the components in the unconfigured state. The assignment step applies only to platforms that support dynamic system domains.

-f

Overrides software state changing constraints. The -f option never overrides fundamental safety and availability constraints of the hardware and operating system.

-l

Lists the state and condition of attachment points specified in the format controlled by the -s, -v, and -a options as specified in cfgadm(1M). The cfgadm_sbd plugin provides specific information in the info field as described below. The format of this information might be altered by the -o parsable option. The parsable info field is composed of the following: cpu

The cpu type displays the following information: cpuid=#[,#.\|.\|.]

Where # is a number, and represents the ID of the CPU. If more than one # is present, this CPU has multiple active virtual processors.

speed=#

Where # is a number and represents the speed of the CPU in MHz.

ecache=#

Where # is a number and represents the size of the ecache in MBytes. If the CPU has multiple active virtual processors, the ecache could either be shared among the virtual processors, or divided between them.

memory

The memory type displays the following information, as appropriate: address=#

Where # is a number, representing the base physical address.

size=#

Where # is a number, representing the size of the memory in KBytes.

permanent=#

Where # is a number, representing the size of permanent memory in KBytes.

unconfigurable

An operating system setting that prevents the memory from being unconfigured.

inter-board-interleave

The board is participating in interleaving with other boards.

source=ap_id

Represents the source attachment point.

target=ap_id

Represents the target attachment point.

deleted=#

Where # is a number, representing the amount of memory that has already been deleted in KBytes.

remaining=#

Where # is a number, representing the amount of memory to be deleted in KBytes.

io

The io type displays the following information: device=path

Represents the physical path to the I/O component.

referenced

The I/O component is referenced.

board

The board type displays the following boolean names. If they are not present, then the opposite applies. assigned

The board is assigned to the domain.

powered-on

The board is powered on.

The same items appear in the info field in a more readable format if the -o parsable option is not specified.
-o parsable

Returns the information in the info field as a boolean name or a set of name=value pairs, separated by a space character. The -o parsable option can be used in conjunction with the -s option. See the cfgadm(1M) man page for more information about the -s option.

-t

Tests the board. Before a board can be connected, it must pass the appropriate level of testing. Use of this option always attempts to test the board, even if it has already passed the appropriate level of testing. Testing is also performed when a -c connect state change function is issued, in which case the test step can be skipped if the board already shows an appropriate level of testing. Thus the -t option can be used to explicitly request that the board be tested.

-x function

Performs an sbd-class function. You can use the following functions: assign

Assigns a board to a domain. The receptacle state must be disconnected or empty. The board must also be listed in the domain available component list. See Dynamic System Domains.

unassign

Unassigns a board from a domain. The receptacle state must be disconnected or empty. The board must also be listed in the domain available component list. See Dynamic System Domains.

poweron

Powers the system board on. The receptacle state must be disconnected.

poweroff

Powers the system board off. The receptacle state must be disconnected.

OPERANDS

The following operands are supported: Receptacle ap_id

For the Sun Fire high-end systems such as the Sun Fire 15K , the receptacle attachment point ID takes the form SBX or IOX, where X equals the slot number. The exact format depends on the platform and typically corresponds to the physical labelling on the machine. See the platform specific information in the NOTES section.

Component ap_id

The component attachment point ID takes the form component_typeX, where component_type equals one of the component types described in "Component Types" and X equals the component number. The component number is a board-relative unit number. The above convention does not apply to memory compontents. Any DR action on a memory attachment point affects all of the memory on the system board.

EXAMPLES

The following examples show user input and system output on a Sun Fire 15K system. User input, specifically references to attachment points and system output might differ on other Sun Fire systems, such as the Sun Fire midrange systems such as the 6800. Refer to the Platform Notes for specific information about using the cfgadm_sbd plugin on non-Sun Fire high-end models.

Example 1 Listing All of the System Board

# cfgadm -a -s "select=class(sbd)"

Ap_Id Type Receptacle Occupant Condition
SB0 CPU connected configured ok
SB0::cpu0 cpu connected configured ok
SB0::memory memory connected configured ok
IO1 HPCI connected configured ok
IO1::pci0 io connected configured ok
IO1::pci1 io connected configured ok
SB2 CPU disconnected unconfigured failed
SB3 CPU disconnected unconfigured unusable
SB4 unknown empty unconfigured unknown

This example demonstrates the mapping of the following conditions:

The board in Slot 2 failed testing.

Slot 3 is unusable; thus, you cannot hot plug a board into that slot.

Example 2 Listing All of the CPUs on the System Board

# cfgadm -a -s "select=class(sbd):type(cpu)"

Ap_Id Type Receptacle Occupant Condition
SB0::cpu0 cpu connected configured ok
SB0::cpu1 cpu connected configured ok
SB0::cpu2 cpu connected configured ok
SB0::cpu3 cpu connected configured ok

Example 3 Displaying the CPU Information Field

# cfgadm -l -s noheadings,cols=info SB0::cpu0

cpuid 16, speed 400 MHz, ecache 8 Mbytes

Example 4 Displaying the CPU Information Field in Parsable Format

# cfgadm -l -s noheadings,cols=info -o parsable SB0::cpu0

cpuid=16 speed=400 ecache=8

Example 5 Displaying the Devices on an I/O Board

# cfgadm -a -s noheadings,cols=ap_id:info -o parsable IO1

IO1 powered-on assigned
IO1::pci0 device=/devices/saf@0/pci@0,2000 referenced
IO1::pci1 device=/devices/saf@0/pci@1,2000 referenced

Example 6 Monitoring an Unconfigure Operation

In the following example, the memory sizes are displayed in Kbytes.

# cfgadm -c unconfigure -y SB0::memory &
# cfgadm -l -s noheadings,cols=info -o parsable SB0::memory SB1::memory

address=0x0 size=2097152 permanent=752592 target=SB1::memory
 deleted=1273680 remaining=823472
address=0x1000000 size=2097152 source=SB0::memory

Example 7 Assigning a Slot to a Domain

# cfgadm -x assign SB2

Example 8 Unassigning a Slot from a Domain

# cfgadm -x unassign SB3
ATTRIBUTES

See attributes(5) for a description of the following attribute:

ATTRIBUTE TYPE ATTRIBUTE VALUE
Stability See below.

The interface stability is evolving. The output stability is unstable.

SEE ALSO

cfgadm(1M), devfsadm(1M), ifconfig(1M), mount(1M), pbind(1M), psradm(1M), psrinfo(1M), config_admin(3CFGADM), attributes(5)

NOTES

This section contains information on how to monitor the progress of a memory delete operation. It also contains platform specific information.

"Memory Delete Monitoring"

The following shell script can be used to monitor the progress of a memory delete operation.

# cfgadm -c unconfigure -y SB0::memory &
# watch_memdel SB0

#!/bin/sh
# This is the watch_memdel script.

if [ -z "$1" ]; then
 printf "usage: %s board_id\en" `basename $0`
 exit 1
fi

board_id=$1

cfgadm_info='cfgadm -s noheadings,cols=info -o parsable'

eval `$cfgadm_info $board_id::memory`

if [ -z "$remaining" ]; then
 echo no memory delete in progress involving $board_id
 exit 0
fi

echo deleting target $target

while true
do
 eval `$cfgadm_info $board_id::memory`

 if [ -n "$remaining" -a "$remaining" -ne 0 ]
 then
 echo $deleted KBytes deleted, $remaining KBytes remaining
 remaining=
 else
 echo memory delete is done
 exit 0
 fi
 sleep 1
done
exit 0
"Sun Enterprise 10000 Platform Notes"

The following syntax is used to refer to Platform Notes attachment points on the Sun Enterprise 10000 system:

board::component

where board refers to the system board; and component refers to the individual component. System boards can range from SB0 (zero) to SB15. A maximum of sixteen system boards are available.

The DR 3.0 model running on a Sun Enterprise 10000 domain supports a limited subset of the functionality provided by the cfgadm_sbd plugin. The only supported operation is to view the status of attachment points in the domain. This corresponds to the -l option and all of its associated options.

Attempting to perform any other operation from the domain will result in an error that states that the operation is not supported. All operations to add or remove a system board must be initiated from the System Service Processor.

"Sun Fire High-End System Platform Notes"

The following syntax is used to refer to attachment points on the Sun Fire high-end systems:

board::component

where board refers to the system board or I/O board; and component refers to the individual component.

Depending on the system's configuration, system boards can range from SB0 (zero) through SB17, and I/O boards can range from IO0 (IO zero) through IO17. (A maximum of eighteen system and I/O boards are available).

The -t and -x options behave differently on the Sun Fire high-end system platforms. The following list describes their behavior: -t

The system controller uses a CPU to test system boards by running LPOST, sequenced by the hpost command. To test I/O boards, the driver starts the testing in response to the -t option, and the test runs automatically without user intervention. The driver unconfigures a CPU and a stretch of contiguous physical memory. Then, it sends a command to the system controller to test the board. The system controller uses the CPU and memory to test the I/O board from inside of a transaction/error cage. You can only use CPUs from system boards (not MCPU boards) to test I/O boards.

-x assign | unassign

In the Sun Fire high-end system administration model, the platform administrator controls the platform hardware through the use of an available component list for each domain. This information is maintained on the system controller. Only the platform administrator can modify the available component list for a domain. The domain administrator is only allowed to assign or unassign a board if it is in the available component list for that domain. The platform administrator does not have this restriction, and can assign or unassign a board even if it is not in the available component list for a domain.

"Sun Fire 15K Component Types"

The following are the names and descriptions of the component types: cpu

CPU

io

I/O device

memory

Memory

Note: An operation on a memory component affects all of the memory components on the board.

"Sun Fire Midrange Systems Platform Notes"

References to attachment points are slightly different on Sun Fire midrange servers such as the 6800, 4810, 4800, and 3800 systems than on the Sun Fire high-end systems. The following syntax is used to refer to attachment points on Sun Fire systems other than the Sun Fire 15K:

N#.board::component

where N# refers to the node; board refers to the system board or I/O board; and component refers to the individual component.

Depending on the system's configuration, system boards can range from SB0 through SB5, and I/O boards can range from IB6 through IB9. (A maximum of six system and four I/O boards are available).

"Sun Fire Midrange System Component Types"

The following are the names and descriptions of the component types: cpu

CPU

pci

I/O device

memory

Memory

Note: An operation on a memory component affects all of the memory components on the board.