I2O_BLOCK_INIT(9) Message-based devices I2O_BLOCK_INIT(9)NAME
i2o_block_init - Block OSM initialization function
Allocate the slab and mempool for request structs, registers i2o_block block device and finally register the Block OSM in the I2O core.
Returns 0 on success or negative error code on failure.
COPYRIGHT Kernel Hackers Manual 3.10 June 2014 I2O_BLOCK_INIT(9)
Check Out this Related Man Page
ri(7) Miscellaneous Information Manual ri(7)NAME
ri - I2O RAID disk interface
bus i2o0 at pci2001 slot 4
controller i2o_bs0 at i2o0 slot 18
The ri driver supports 32-bit and 64-bit PCI backplane RAID controllers with Intelligent I/O (I2O) host interface supporting backend SCSI
channels. I2O is an open architecture for developing device drivers that is independent of the operating system, processor platform, and
system I/O bus. The communication model is based on a message-passing protocol reducing the number of driver interfaces needed. A single
driver called an Operating System Module(OSM) is provided to support all Block Storage class devices. Any RAID system that supports the
I2O interface functions with this driver. These devices are designated as ri to distinguish them from ra, re, and rz (SCSI) devices.
The StorageWorks Command Console Utility (SWCC) is supported for online configuration and monitoring of I2O raid devices.
The following rules are used to determine the major and minor numbers that are associated with an ri type disk. A dynamic major number is
used to represent ri block devices, and is assigned during initial configuration. The current major number can be determined using the
following command: # /sbin/devswmgr -getnum i2o_bs The preceding command displays the driver switch reservation list, showing the device
instance and major number for the i2o_bs driver. An alternative method is: # /sbin/sysconfig -q i2o_bs
The highest 14 bits of the 20-bit minor number represents a particular Block Storage device, while lowest six bits of the minor number
specify disk partitions (identified by the letters, a through h) on that device.
The device special file names associated with ri disks are based on conventions that are closely associated with the minor number assigned
to the disk. The standard device names begin with ri for block special files and rri for character (raw) special files. Following the ri
is the unit number and then a letter, a through h, that represents the partition. Throughout this reference page, the question mark (?)
character represents the unit number in the name of the device special file. For example, ri?b could represent ri0b, ri1b, and so on.
A disk can be accessed through either the block special file or the character special file. The block special file accesses the disk using
the file system's normal buffering mechanism. Reads and writes to the block special file can specify any size. This capability avoids the
need to limit data transfers to the size of physical disk records and to calculate offsets within disk records. The file system can break
up large read and write requests into smaller fixed size transfers to the disk.
The character special file provides a raw interface that allows for direct transmission between the disk and the user's read or write buf-
fer. A single read/write operation to the raw interface results in exactly one I/O operation. Consequently, raw I/O may be considerably
more efficient for large transfers.
For systems with ri disks, the first software boot after the system is powered on may take longer than expected. This delay is normal and
is caused by the software spinning up the ri disks.
The ri driver handles all disk drives that can be connected to the I2O RAID controller. To determine which drives are supported for spe-
cific CPU types and hardware configurations, see the hardware installation and configuration information for your I2O system.
I2O RAID Controllers are viewed in all cases as RI type disks. There are some notable differences that should be taken into consideration
when configuring a RAID device: Currently only sector sizes of 512 bytes are supported. Logical Volume sizes are not fixed sizes as com-
pared to other disk devices. The size of the Logical Volume is configurable based on needs. The dynamic nature of Logical Volume sizes is
dealt with by defining RAID devices as DYNAMIC. Only partitions a, b, c, and g are defined. If necessary, the disklabel(8) command can be
run to change and define partitions for RAID devices
Usually, the ri?a partition is used for the root file system and the ri?b partition as a paging area. The ri?c partition can be used for
disk-to-disk copying because it maps the entire disk.
The starting location and length (in 512 byte sectors) of the disk partitions of each drive are shown in the following table. Partition
sizes can be changed by using the disklabel(8) command.
I2O RAID partitions for systems
based on the Alpha AXP architecture
disk start length
ri?a 0 131072 ri?b 131072 262144 ri?c 0 end of media ri?d 0 0 ri?e 0 0 ri?f 0 0
ri?g 393216 end of media ri?h 0 0
RELATED INFORMATION RAID(7), SCSI(7), tz(7), rz(7),disklabel(8), MAKEDEV(8), uerf(8) delim off