IRIX Advanced Site and Server Administration Guide

Appendix C
IRIX Kernel Error Messages

This appendix lists error messages you may receive from the IRIX operating system. The operating system handles error conditions according to their degree of severity and divides the error messages into three classes: NOTICE, WARNING, and PANIC.

NOTICE messages provide information on the system status. These messages can sometimes help you anticipate problems before trouble occurs. See "NOTICE Messages".



WARNING messages indicate that the system might stop functioning unless you take corrective action. See "WARNING Messages".



PANIC messages indicate that a problem is severe enough that the system comes to a halt. The cause is usually a hardware problem or a problem in the kernel software. This type of message occurs occasionally; it should not cause much concern. If a particular PANIC message occurs repeatedly or predictably, however, contact your service representative. See "PANIC Messages".

The error messages in this appendix are listed alphabetically in their respective classes. The messages that follow relate to tunable parameters. For other error messages, see the file /usr/include/sys/errno.h, the intro(2) reference page, and the Owner's Guide for your system.

NOTICE Messages

diskname: Swap out failed blkno x (page still in memory)

A swap write failed. No action is necessary as the page is still in memory.



Insufficient memory to lock/allocate # pages

There is not enough memory available to open the requested pages.



system call failed for process name

There is insufficient free memory to execute the system call; reduce the system load and try again, or add more memory.



"str" - swpuse count overflow

More than 256 processes are sharing the same page of swap. A copy has been made. No action is required.



useracc - no memory to lock page

There is insufficient free memory to lock a user data page in memory to service a read or write system call to a raw device. Reduce the system load, reduce the size of the raw I/O buffer in the user program, or add more memory.

WARNING Messages

iget - inode table overflow

Each active file in the system requires an entry in the inode table, and this table has overflowed. Reduce the system load. If this happens frequently, increase the tunable parameter ninode using systune(1M) and follow any directions given by that utility to activate the change.



mfree map overflow

Fragmentation of some resource (such as message queues) contributed to the loss of some of the resource. No action necessary.



out of physical memory, nbytes=#

No more physical pages are available. Reduce the system load or add more memory.



Paging Daemon (vhand) not running - NFS server down?

The system determines that vhand is not executing, possibly because it is waiting for an I/O transfer to complete to an NFS server (especially if the NFS file system is hand mounted). No action should be necessary as the system will restart vhand when needed.

PANIC Messages

bumpcnt - region count list overflow

Indicates an unresolvable problem with the operating system. Reboot your system.



getpages - pbremove

Indicates an unresolvable problem with the operating system. Reboot your system.



Ran out of action blocks

A resource used by the multiprocessor kernel for inter-CPU interrupts has run out. If this happens frequently, use the systune(1M) command to increase the value of the parameter nactions.



vfault - bad dbd_type

The page being faulted in is not a recognized type: either demand fill, demand zero, in file or on swap (operating system error). Reboot your system and if the error persists, check your application and your disk.

Copyright © 1997, Silicon Graphics, Inc. All Rights Reserved. Trademark Information