The following device management features and enhancements have been added to the Solaris 10 11/06 release.
This Solaris release provides support for the PCI Express (PCIe) interconnect for both SPARC and x86 systems.
PCIe is designed to connect peripheral devices to desktop, enterprise, mobile, communication, and embedded applications.
The PCIe interconnect is an industry-standard, high-performance, serial I/O bus.
The PCIe software provides the following features in this Solaris release:
Support for extended PCIe configuration space
Support for PCIe baseline error handling and MSI interrupts
Modified IEEE-1275 properties for PCIe devices
PCIe hot-plug support (both native and ACPI-based) by enhancing the cfgadm_pci component of the cfgadm command
ATTN button usage-based PCIe peripheral autoconfiguration
The following cfgadm example output displays the hot-pluggable PCIe devices on x86 systems. Note that the following display might differ from platform to platform. Check your hardware platform guide for the correct cfgadm syntax.
# cfgadm pci Ap_Id Type Receptacle Occupant Condition pcie1 unknown empty unconfigured unknown pcie2 unknown empty unconfigured unknown pcie3 unknown empty unconfigured unknown pcie4 etherne/hp connected configured ok pcie5 pci-pci/hp connected configured ok pcie6 unknown disconnected unconfigured unknown
The administrative model for hot-plugging PCIe peripherals is the same as for PCI peripherals, which use the cfgadm command.
For more information, see the cfgadm_pci(1M) man page and System Administration Guide: Devices and File Systems. Check your hardware platform guide to ensure that PCIe and PCIe hot-plug support is provided on your system. In addition, carefully review the instructions for physically inserting or removing adapters on your system and the semantics of device autoconfiguration, if applicable.
For more information about PCIe technology, see http://www.pcisig.com.
A new Fault Management Architecture-based diagnosis engine (DE) is provided on the Sun Fire X4500. This DE monitors the disk drives for predictive failures by using the SMART technology in the disk drive's own firmware. When a disk failure is imminent, the LED next to the disk is illuminated and a Fault Management Architecture fault is generated. This fault alerts the administrator to take specific action to ensure system availability and full performance.
Ipge drivers are used in Ontario and other SPARC platforms that have the NorthStar card installed. E1000g drivers are used in all other platforms.
Starting with this release, Ontario and other SPARC based platforms transition from ipge to e1000g drivers. This feature makes the e1000g the default driver for all Sun platforms that use Intel 1G chipsets. With the transition, the customer does not need to know which platform the ipge or the e1000g driver covers or which driver to install in a particular platform. This feature reduces system management complexity.
For more information, see “Certain 3rd Party Applications May Break on Transition From ipge to e1000g Network Driver” on http://sunsolve.sun.com/.
The Solaris fibre channel logical unit number (LUN) masking feature enables system administrators to prevent the kernel from creating device nodes for specific unapproved LUNs.
For more information, see the fp(7d) man page.
Extended Message Signaled Interrupts (MSI-X) are an enhanced version of MSI interrupts. With MSI-X support, device driver writers have a choice between MSI and MSI-X interrupts. MSI-X interrupts are now supported on SPARC PCI-Express platforms (Ultra 45 and Sun Fire T2000). The Sun Fire T2000 might also include the Sun Fire T1000 machine.
A new mdb/kmdb debugger command, ::interrupts, is also provided to retrieve a device's registered interrupt information on supported SPARC and x86 systems.
For more information, see “Interrupt Handlers” in Writing Device Drivers.
The following utilities have been enhanced to detect when a specified device is in use:
mkfs and newfs
These enhancements mean that these utilities might detect some of the following usage scenarios:
Device is part of a ZFS storage pool
Device is a dump or swap device
Mounted file system or an entry for the device exists in the /etc/vfstab file
Device is part of a live upgrade configuration
Device is part of a Solaris Volume Manager configuration or Veritas Volume Manager configuration
For example, if you attempt to use the format utility to access an active device, you will see a message similar to the following:
# format . . . Specify disk (enter its number): 1 selecting c0t1d0 [disk formatted] Warning: Current Disk has mounted partitions. /dev/dsk/c0t1d0s0 is currently mounted on /. Please see umount(1M). /dev/dsk/c0t1d0s1 is currently used by swap. Please see swap(1M).
However, these utilities do not detect all scenarios in the same way. For example, you can use the newfs command to create a new file system on a device in a live upgrade configuration. You cannot use the newfs command to create a new file system on a device that is part of a live upgrade configuration if it also has a mounted file system.