HP Tru64 UNIX and TruCluster Server Version 5.1.B-4 Patch Summary and Release Notes (13156)

• Addresses a problem in IPv6 subsystem that causes a system to panic.
• Adds support for AlphaServer ES47/ES80/GS1280 platforms to allow the use of processors
having different speeds in the same chassis.
• Corrects the problem of the statfs() function returning EINVAL when operating on an
fattach() function clone streams device.
• Addresses a scaling issue seen on large multiprocessor systems in dealing with the class
scheduling subsystem.
• Fixes a problem in gh_chunks allocation on some configurations.
• Improves file caching performance for large files on NUMA systems.
• Fixes panic: simple_lock_terminate: lock busy.
• Fixes cluster crashing when VMAC is enabled
• Fixes an underlying problem in the NFS client that can lead to a panic on a single system or
an assertion failure panic on a cluster.
• Fixes the number of cylinders defined in the last cylinder group of a UFS file system that
has been extended.
• Fixes various problems in the bcm driver for DEGXA Gigabit Ethernet adapters.
• Removes the 32-bit block address restriction in SpecFS read and write.
• Removes excess data stored in the auditing of the swapctl syscall.
• Adjusts the automatic computation of the size of the NFS server's duplicate request cache
to accommodate only active RADs on a multi-RAD system.
• Increases ability to audit memory wire and unwire operations.
• Resolves a problem affecting IPv6 sockets where a packet transmission might fail when the
destination address is not specified.
• Fixes a kernel memory fault panic while running the IDRIS benchmark.
• Corrects a kernel memory fault caused by configuring an IPv6 address on a legacy network
interface card.
• Fixes multiple problems with IPv6 advanced API implementation.
• Corrects a problem in which calling aio_write with a negative size may cause a system panic.
• Corrects a potential security vulnerability reported by SSRT2323.
• Fixes a leak of kernel address space.
• Corrects a problem in which a read or write operation to a changer device creates an
unkillable process.
• Corrects the panic “trap: invalid memory read access from kernel mode in procfs_psinfo().”
• Corrects a condition in which big page memory allocations on NUMA systems may allocate
remote memory too soon.
• Fixes a problem in which clua.mod does not handle TCP RST messages appropriately.
• Updates ICMP code to add redirect timeout support and nonmodifiable route table entries.
• Addresses performance issues with the select() function.
• Accelerates the booting of memory GS systems.
• Adds pfilt_loopback and pfilt_physaddr kernel flags for controlling packetfilter written
packets.
• Corrects a "simple_lock: time limit exceeded" issue that can occur in the CAM I/O subsystem
when MCS locking is disabled.
• Provides support for the dynamically loadable packet filter.
• Fixes IP multicast packets to work with loopback traffic
• Improves NFS client performance on NUMA systems by doing a better job distributing data,
locks, and nfsiod threads across Resource Affinity Domains (RADs).
• Enables suitable monitoring of individual NetRAIN interfaces
96 Tru64 UNIX Patches