home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
IRIX Base Documentation 1998 November
/
IRIX 6.5.2 Base Documentation November 1998.img
/
usr
/
relnotes
/
eoe
/
ch4.z
/
ch4
Wrap
Text File
|
1998-11-02
|
49KB
|
1,783 lines
- 1 -
4. _B_u_g__F_i_x_e_s
This section of the release notes lists bugs that have been
fixed in IRIX 6.2 through IRIX 6.5 and in IRIX Maintenance
Release. Bug numbers from the Silicon Graphics bug tracking
system are included for reference.
4.1 _I_n_s_t_a_l_l_a_t_i_o_n__t_o_o_l_s
4.1.1 _B_u_g_s__f_i_x_e_d__i_n__I_R_I_X__6_._2__t_h_r_o_u_g_h__I_R_I_X__6_._5
+o 631728: glReadPixels incorrect with short color index
data
+o 592548 : write-protected MOD -- device stayed write-
protected after protection was disabled
+o 566235: 6.3 O2 GL_FASTEST depth-cueing problem
+o 560428: Inst or swmgr would sometimes calculate the
volume header device name incorrectly causing exitops
that use $vhdev to fail.
+o 544319: Inst or swmgr would occasionally core dump
when installing over a failed NFS mount.
+o 544082: Swmgr would hang under certain rare conditions
after a right click on a product name.
+o 542709: Remote installs using swmgr or inst could
leave unnecessary shell processes running on the
server.
+o 542322, 542301, 470457: Selection files without a from
line failed when using inst or swmgr. Selection files
consisting solely of remove commands failed. Selection
files with remove commands and other commands always
executed the remove commands even if the optional
version number did not match. Selection files did not
honor the open command.
+o 535947: Inst or swmgr product rules could get
corrupted under certain rare cirmcumstances during
network installs.
+o 508209: Patches that change the base file type (ie.
dir to symlink) would not install properly.
+o 505714: Inst or swmgr now allows multiple products of
the same name and different versions to be loaded in
the same installation session.
- 2 -
+o 487549 : 6.5.1m mailbox fix makes inside black (was
mailbox -S cannot see text on the O2)
+o 486395, 369473: The logic to obtain the netmask
setting for the miniroot was improved.
+o 485269 Inst list command doesn't sort by product name
correctly.
+o 459827: Inst history file checksums were not being
updated properly after rqs processing.
+o 456982: Using the get info capability of swmgr on a
file could cause a core dump.
+o 450734: Starting up swmgr from a remote toolchest did
not display correctly when a root password was present.
+o 448628 #448586 #448583 #448447 Swmgr had problems
displaying mulitbyte text.
+o 447490: Swmgr would not start up if it didn't get the
colors it wanted.
+o 241982: Swmgr autoselect preference was being ignored.
4.1.2 _B_u_g_s__f_i_x_e_d__i_n__I_R_I_X__6_._5_._1__a_n_d__I_R_I_X__6_._5_._2
+o 527125: Inst would dump core if LANG env variable was
set to ja_JP.EUC.
+o 560351: Warn user when 'versions removehist' is used.
+o 562156: Swmgr didn't obey -A or -u all.
+o 568120: Inst conflict command sometimes produced
garbled output.
+o 579670 598339: Inst history not updated after
cancelled/failed rqs.
+o 579966: Inst/Swmgr could generate a conflict that was
not resolvable.
+o 586052: Inst/Swmgr would put user into interrupt menu
unnecessarily on patch removal when a directory was
missing.
+o 591919: Provide a more helpful message from postinst
after an autoconfig failure.
- 3 -
+o 595429: Showprods could destroy a ".tardist" file.
+o 599626: Inst core dumps using selection file
containing "bad" remote hostname.
+o 607421: Remote CD installs would fail if CD did not
contain ".IM" file.
+o 617424: Diskless installs of overlays did not work.
+o 619380: Major inst/swmgr performance problem when
installing overlays.
+o 622317: Using conflicts command in inst -c command
script would terminate script.
+o 633238: Swmgr would core dump during some overlay
installations.
+o 634528: Swmgr and Inst would not install some files if
product name != filename
4.2 _C_o_m_m_a_n_d_s
4.2.1 _B_u_g_s__f_i_x_e_d__i_n__I_R_I_X__6_._5_._2
+o 557237: chkutent infinite loop with "utmpx dup - ...
22126 - see chkutent(1M)"
+o 578656: touch command incorrectly processes -amc
options
+o 581266: cmp reads skip arguments as 32-bit values
+o 585205: "who -u" give negative numbers for PID
+o 591632, 623315: login from console fails
+o 598369: /sbin/ioconfig invoked with insufficient
privilege during boot
+o 604957: chlabel command does not bypass DAC when needed
+o 606798: /usr/sbin/showprods Y2K problem - year now
displayed in full
+o 611564: Kudzu irix6.5 Compiler problem with
COMPILER_DEFAULTS_PATH.
+o 614548: bstream command fails at 2GB point, and above.
- 4 -
+o 618225: xterm does not open window while user is not a
privileged user
+o 617321: /bin/at security vulnernability.
+o 618293: cpr of rcp can cause rcp to hang in 6.4, 6.5
+o 624271: usr/sbin/cview dumps core (eoe.sw.cpr)
+o 622784: Expr core dumps
+o 622059: unlimit without arguments returns error
+o 630487: more "+" option functionality broken
+o 614578: BSH: anything linked to builtin_exec forks
itself if argument is resident
+o 607048: /bin/sh (ksh) dumps core with really long argv
list.
+o 590776: ksh: Command substitution $( ... ) works
incorrectly
+o 577134: ksh: == operator not supported
+o 546952: AR and ARFLAGS not overridden by environment
variables
+o 542135: ksh goes wild on symbolic links use as
includes
+o 524930: sort does not work with subfields
4.2.2 _B_u_g_s__f_i_x_e_d__i_n__I_R_I_X__6_._5_._1
+o 535033: String abbreviations of "tr" command aren't
working in Japanese env.
+o 560572: userenv dumps core when it can't read files
+o 562153: man display japanese data incorrectly
+o 577494: syslog gives incorrect facility when priority
panic is used
+o 578996: ICRASH should look for both PROM and eframe
formatted register values when analyzing a system core
dump generated by an NMI.
- 5 -
+o 580871: rdist via script hangs
+o 583824: Hungarian locale (LC_TIME) problem
+o 592248: rdist core dumps when interrupted
+o 594441: ksh fails to execute correct binary when full
path is given
+o 597464: log rotation cron job failures.
+o 599488: more doesn't always get the right input mode
(This fix contains a workaround in the more command. It
reduces the problem. But it doesn't contain a fix on
the tty side yet.)
+o 607447: mailx trashes large email files
+o 620266: fccli core dump
4.3 _N_e_t_w_o_r_k_i_n_g
4.3.1 _B_u_g_s__f_i_x_e_d__i_n__I_R_I_X__6_._5_._2
+o 507396: Rare Origin and OCTANE prom failure while doing
bootp that is triggered only if there is a simultaneous
1500 byte broadcast packet sequence from another
machine on the same ethernet.
+o 579321: Need to use netgroups from local file without
NIS running
+o 581349: jan18r.G: NFS locking: EDEADLK not returned
from fcntl()
+o 585346: jan18r.G: NFS rename() errno problem for Posix
+o 586048: jan18r.G: NFS: VSX+XNFS test failure for
Network Lock Manager
+o 587189: umount -k fails on xfs file system with file
system busy
+o 591776: Rename on full non-root filesystem paniced
system
+o 590099: NFSv3 file truncation can fail
+o 597590: (LANL) NFS V3 mount fail with : can't decode
result
- 6 -
+o 601220: rpc.passwd does not honor existing permissions
+o 611007: NFS v2 open O_CREAT fails on existing file
+o 614246: Parse0 ruleset missing, breaks 2231 anti-spam
feature
+o 616890: NFSv2 WRITE failures caused by client sending
blocks > 8192
+o 617435 - Sendmail make usernames case sensitive
+o 617536: poor NFS write performance when file locks are
held
+o 619650: compress does not preserve timestamp info over
NFS filesystems
+o 621001: TCP stack not leaving enough room for HIPPI
headers and causing extra memory allocation work in the
HIPPI device driver.
+o 621106: autofs does not use UNS (nsd) to lookup maps
+o 625699: nsd should complain if ypserv=on and system has
not been ypinit(ed)
+o 623617: exclusive create call on sgid set directory
fails for some users
+o 626331: vnode reference count leaks were introduced
+o 627767: inetd could be killed by port-scan attacks.
+o 629253: gethostbyname() failure with unloaded nsd cache
+o 630393: nsd does not fallback to using domain from a
fully qualified hostname as a default DNS domain.
+o 630483: nsd ignores > 1 netgroups in passwd file
+o 631209: interrupted fsync to NFSv2 file always fails on
retry
+o 632359: /var/yp/mdbm_parse does not correctly push the
netgroup.byhost map
+o 633284: nsadmin clean fails to clear old entries from
cache
- 7 -
+o 633344: NSD causes NIS+ servers in NIS-compatibility
mode to fill SYSLOG with useless messages
+o 633389: proclaim needs to process gateway and static
route options
+o 633902: "nsd -a nis_servers=" option does not work for
down server in list
4.3.2 _B_u_g_s__f_i_x_e_d__i_n__I_R_I_X__6_._5_._1
+o 566288: Autofsd fails on entries > 1023 characters
+o 571213: inconsistent locking for r_flags field in
struct rnode
+o 571214: NFS TCP starvation
+o 573530: netinet/tcp.h can be included more than once
+o 573604: mountd fails replying to automount -hosts with
large exports list
+o 574188: snetd fails with 3 interfaces
+o 577651: autofs and automount requests to non-SGI
servers fail AUTH_TOOWEAK
+o 580372: ipfilterd interferes with NFS
+o 580878: NFS file access times don't change
+o 581230: NFSv2 versus NFSv3 pathconf() failure
+o 581280: NFS file locking: bad errno with fcntl
+o 581816: NFSv2 versus NFSv3 utime() failure
+o 582783: tcp don't output good enough?
+o 585256: PANIC: CPU 1: Ran out of action blocks
+o 585359: NFS write() failure when file size limit not
exceeded
+o 586219: excessive zone memory uszge by cachefs
+o 586244: ls of NFSv3 directories is incomplete
+o 586988: "NFS write error 13 on host" errors with patch
2656
- 8 -
+o 587590: Extra call to SMAP_RELE allows socket to be
closed prematurely, leading to PANIC
+o 588290: rpc.statd thrashing filesystems
+o 589176: VSX+XNFS: symbolic link returns incorrect
status
+o 589199: VSX+XNFS: readlink returns incorrect status
+o 591786: loopback NFS mounts subject to deadlock
+o 593276: file truncation seen on NFS mounted filesystem
+o 594841: Regression: mount_dos can't mount hdf file
+o 597414: apr12e: kernel fault from cachefs_lookup()
+o 598168: kernel rpc doesn't understand SAMP
+o 598180: autofs doesn't work in Trix
+o 599777: BDS Regression in kudzu 6.5 from 6.2/6.3/6.4.
+o 600574: remote mount of CDROM doesn't show all files
+o 601086: nfs_sched() locking is broken
+o 601321: 5.3 "at" patch number 2884 is BROKEN.
+o 601674: cachefs causes panic on o2 r10k
+o 602702: satmp_init_reply() doesn't handle 64-bit arg
correctly
+o 603283: APP: SDRC: Kernel Panic - apr12f
+o 604040: autofs managed directories disappear
occasionally
+o 604288: NFS write timeouts due to extra UDP garbage
+o 608319: 'wakeupat' fails to handle leap years.
+o 608653: poor tcp packetization leads to extraneous
bcopys
+o 608660: nfs3_rah variable should be systuneable
- 9 -
4.3.3 _B_u_g_s__f_i_x_e_d__i_n__I_R_I_X__6_._2__t_h_r_o_u_g_h__I_R_I_X__6_._5
+o 8180: YP should use multicast (IRIX 6.2)
+o 13146: create an empty /etc/init.d/network.local as
default file (IRIX 6.2)
+o 232115: static route for /etc/init.d/network startup
script (IRIX 6.2)
+o 254795: panic in tpitcp_function (IRIX 6.2, 6.4)
+o 259508: TLI program does not run on all SGI platforms
(IRIX 6.2, 6.3, 6.4)
+o 273287: Account expiration via /etc/shadow does not
stop XDM/FTP logins. (IRIX 6.2)
+o 284409: tpisocket race conditions cause kernel to panic
(IRIX 6.2)
+o 291184: netstat -A reportes UNIX domain socket types
incorrectly (IRIX 6.2)
+o 294178: named needs to support round-robin CNAME
aliases (IRIX 6.2)
+o 310295: simultaneous open of named stream pipe hangs
processes (IRIX 6.2)
+o 32332: ypserv case sensitive (IRIX 6.2)
+o 323866: Patch 797 (IPAliases portion) shows strange
routing table with ppp setups
+o 327564: strange pauses outputting to tty/stream (IRIX
6.2, 6.3, 6.4)
+o 331622: BUG - mrouted keeps complaining about a
duplicate prune (IRIX 6.2, 6.3)
+o 334558: IRIX 5.3 telnet loses 4096th character. (IRIX
6.2)
+o 344692: gated doesn't install proper subnet routes
(IRIX 6.2)
+o 346471: strioctl using wrong lock in TCSETLABEL case
(IRIX 6.2)
- 10 -
+o 348335: rpcbind does not work with TOT kernels (IRIX
6.2)
+o 348619: arp -a doesn't output al arp table entres.
(IRIX 6.2)
+o 351245: /etc/init.d/network extension for static routes
(IRIX 6.2)
+o 351375: ipaliases on primary causes inetd to segv on
startup (IRIX 6.2)
+o 351600: When IP aliases are deleted, their host routes
sometimes aren't (IRIX 6.2
+o 352269: routing socket get-interfaces needs to produce
aliases (IRIX 6.2, 6.3, 6.4)
+o 353160: routing broken for PPP links with not all 1's
netmask
+o 353174: cannot add multicast groups to PPP links
+o 353649: ftpd's hanging around after other side has gone
away (IRIX 6.2)
+o 354880: del_vif() can return while holding mroute lock
+o 360200: Badly-behaved web clients can leave connections
in closing on server
+o 360309: kernel panic in tcp_fasttimo at NTT
+o 360697: Multicast rate limiting does not use MP locks
+o 361330: Increase IP_MAX_MEMBERSHIPS to > 20 ( say 300 )
(IRIX 6.2)
+o 362346: fixes need for rsvp on machines running mrouted
(IRIX 6.2)
+o 363009: invalid mbuf causing tblmiss in m_freem (IRIX
6.2)
+o 363058: timeslave gets killed on IR / 6.2 MR (IRIX 6.2)
+o 364172: ipaliases causes problems with .rhosts (IRIX
6.2)
+o 364653: ip-alias kernel panic (IRIX 6.2)
- 11 -
+o 365048: MP deadlock in IP multicast code
+o 366431: KERNEL FAULT when arp table size is 328 (IRIX
6.2)
+o 368408: TCP resets on hardware checksummed boards don't
work (IRIX 6.2)
+o 368997: inetd leaves socket open when service is
shutdown (IRIX 6.2)
+o 370077: MP deadlock in hash table code (IRIX 6.2)
+o 372087: netstat -f unix reports nothing on 32-bit
systems (IRIX 6.2)
+o 373847: timeslave WWV/Traconex problems (IRIX 6.2)
+o 374988: IRIX 6.2 crashes with unused network interface
and new routed (IRIX 6.2)
+o 375099: ifconfig in patch 797 (and successor 1092) does
not set destination address
+o 375172: KERNEL FAULT
+o 375298: ifa->ifa_start_inifaddr addr null (IRIX 6.2)
+o 375298: ip-aliasing bug (IRIX 6.2)
+o 376741: MP systems can crash in accept() if conn.
terminated prematurely
+o 376976: spinunlock_pmonitor kernel segv under heavy ftp
load (IRIX 6.2)
+o 377322: eplex interfaces appear to get stuck (IRIX 6.2)
+o 379703: crash in in_pcbnotify() (IRIX 6.2)
+o 380174: IP forwarding far too slow in IRIX 6.2
+o 382081: buglet in netstat -C on SIGCONT (IRIX 6.2)
+o 382134: netstat dumps core in -C screen 8 (IRIX 6.2)
+o 387519: soclose() can leave sockets locked if soabort()
fails
+o 389819: select() race when accepting new connections
(IRIX 6.2)
- 12 -
+o 389824: sosend() can leave socket locked (IRIX 6.2)
+o 390346: Customer believe that incorrect state is
returned (IRIX 6.2, 6.3, 6.4).3, 6.4).3, 6.4).3,
6.4).3, 6.4)
+o 391121: tli program crashes/hangs 6.2 machines. (IRIX
6.2, 6.4)
+o 391952: ipfilterd man page in wrong ISM (IRIX 6.2)
+o 393088: /usr/sbin/ProclaimServerMgr couldn't stat
"/var/dhcp/etherToIP" (IRIX 6.2)
+o 394059: bootp doesn't work through IP aliases on 5.3
and 6.2 (IRIX 6.2)
+o 394367: named reply with primary ip instead of alias ip
(IRIX 6.2)
+o 394867: ICMP unreachables cause multicast havoc
+o 396323: inetd seg-fault looking up invalid password
entry (IRIX 6.2)
+o 396407: memory leak in error handling in stream open
(IRIX 6.2)
+o 397235: ypbind and ypbind assumes a system has at most
20 interfaces or IP addresses
+o 399569: program sleeping at sbunlock_wait, not killable
(IRIX 6.2, 6.4)
+o 404031: rdist has overflow in subroutine call stack.
(IRIX 6.2, 6.3, 6.4)
+o 406579: ftpd core dump in passive because of invalid
uid (IRIX 6.2)
+o 408070: with patch 1247 installed, broadcast packets
are not looped back (IRIX 6.2)
+o 409947: `route add 255.255.255.255 10.1` fails & `route
get 255.255.255.255` crashes (IRIX 6.2)
+o 415809: Cosmo MediaBase sends multicast packets too
large for rate limiting (IRIX 6.2)
+o 416381: SYN_RCVD connections causing tcp ports to hang.
(IRIX 6.2, 6.3, 6.4)
- 13 -
+o 417545: named error: Server Failed (IRIX 6.2)
+o 418059: Bad data is causing a DNS melt-down with bind
4.9.3 (IRIX 6.2)
+o 421674: sockets left locked in tpisocket.c
+o 422437: mbuf exhaustion when heavy ATM LANE traffic
occurs (IRIX 6.2)
+o 422441: in_pcbnotify() can be very expensive (IRIX 6.4)
+o 423124: sleep loop in tcp_close() can hang network
(IRIX 6.2)
+o 427349: autopush fails on 64 bit Irix 6.2 (IRIX 6.2)
+o 427433: SYN bombing (IRIX 6.2, 6.3, 6.4)
+o 428441: getsockopt() of unknown value causes double
free of mbuf (IRIX 6.2)
+o 428841: I/STHREAD FAULT (IRIX 6.4)
+o 429599: arp_lock and IFNET_LOCK() ab<->ba hang (IRIX
6.4)
+o 430154: snoop sockets broken
+o 430549: ipfilter panics on Challenge L (IRIX 6.2)
+o 432928: xwsh seems to set terminal modes incorrectly
(IRIX 6.2, IRIX 6.4)
+o 433439: UDP & SO_REUSEADDR not working right on 6.2?
(IRIX 6.2)
+o 434797: fifo lock problem (IRIX 6.2)
+o 438125: ping can crash a system (IRIX 6.2)
+o 438858: race in tpisocket tests
+o 439590: streams perf problem (IRIX 6.2)
+o 439983: invalid kptbl entry
+o 440572: svr4net still broken on ficus-ssg (IRIX 6.2,
IRIX 6.4)
- 14 -
+o 441987: double trip on socket lock
+o 443482: System paniced when attempting to connect over
socket.
+o 444215: furnace panic'ed in tpisockd
+o 446660: kernel crash during cots/listen X/Open VST test
run
+o 446662: kernel crash during cots/snddis X/Open VST test
run
+o 450354: mbuf leak while running routed or other
SO_PASSIFNAME users
+o 450482: tcp throughput issues on dirty network. (IRIX
6.4)
+o 453788: ppp can't use '#' in username, or any other
field (IRIX 6.2, 6.3, 6.4)
+o 453965: processes can hang in exit() due to negative
tty reference count (IRIX 6.2, 6.4)
+o 454697: /usr/include/netinet/in_var.h - undefined
member and incomplete type problems (IRIX 6.2)
+o 455121: PANIC in streams service routines (IRIX 6.2)
+o 455647: routed complains about xpi0 duplicate addresses
(6.4) (IRIX 6.2)
+o 456853: wrong length in kernel-generated routing socket
messages (IRIX 6.2)
+o 458244: keepalive timer should kill connections in
FIN-WAIT-2 (IRIX 6.2, 6.3, 6.4)
+o 459724: tpisocket fails to set message types correctly
on replies (IRIX 6.2, 6.3, 6.4)
+o 459895: inetd sometimes fails to find user "root" (IRIX
6.2, 6.3, 6.4)
+o 459936: security issue with talkd (IRIX 6.2, IRIX 6.3,
IRIX 6.4)
+o 460854: 6.2 ipfilterd when filtering over vfe will
cause system to hang.
- 15 -
+o 461415: sendto() fails w/o ethernet cable (IRIX 6.3)
+o 462312: problem with user-level appletalk on 6.2 (IRIX
6.2, 6.4)
+o 463113: DHCP Server may give out duplicate addresses if
range exhausted (IRIX 6.2, 6.3, 6.4)
+o 463119: DHCP server does not correctly work with
alternate hosts and ethers file. (IRIX 6.2, 6.3, 6.4)
+o 463120: DHCP server does not allow preassignment of dns
addresses. (IRIX 6.2, 6.3, 6.4)
+o 466105: ethernet driver on O2 systems has mbuf leak
(IRIX 6.3)
+o 466425: O2 6.3 for R10k panic in ether_output (IRIX
6.3)
+o 467421: potential structure leak in tpisocket
+o 468607: bookkeeping error in ip_reass() (IRIX 6.2, 6.3,
6.4)
+o 469183: dhcp_relay coredumps under some conditions
(IRIX 6.2, 6.3, 6.4)
+o 469471: kernel mbuf allocator does not scale (IRIX 6.4)
+o 470658: Omega launch of frame4 breaks with patch 1572
(IRIX 6.2, 6.4)
+o 470827: dhcp_bootp core dumps (IRIX 6.2, 6.3, 6.4)
+o 470993: network input does not scale (IRIX 6.4)
+o 471851: ef driver performance improvements (IRIX 6.4)
+o 472592: accept crashes system (IRIX 6.4)
+o 473007: tpisocket can reference invalid socket pointer
(IRIX 6.4)
+o 473346: Crossed SYNs defense not quite right (IRIX 6.2,
6.3, 6.4)
+o 473347: 10% CPU reduction possible in tcp_zaptw() (IRIX
6.4)
- 16 -
+o 473348: sosend() can be re-done to avoid using bcopy()
as much (IRIX 6.4)
+o 473349: several small bugs in TCP make it hard to
sustain high conns/sec (IRIX 6.4)
+o 473385: IP_HDRINCL does not work on IP options and
ip_id is bad (IRIX 6.2, 6.3, 6.4)
+o 473525: Ipfilterd Hangs Origin 200 server on
IMMEDIATELY (IRIX 6.4)
+o 474870: 32-bit I_FDINSERT doesn't work right on 64-bit
kernel (IRIX 6.4)
+o 477568: iflist sysctl does not return unconfigured
interfaces (IRIX 6.2, 6.3, 6.4)
+o 477707: Loading configurations that are disabled causes
incorrect behaviour.
+o 479699: jabber detected error (auto-negotiation fails)
(IRIX 6.3)
+o 480545: T_DISCON_REQ not processed correctly when in
state T_WRES_CIND
+o 480831: IRIX 6.2 and newer have different interface
ordering than 5.3
+o 481297: Server gives out address not in rage with -x
option
+o 481873: Decrypted password left in core dump after a
long argument is given to list (IRIX 6.2, 6.3, 6.4)
+o 481934: O2 mace ethernet does not work at 10Mb if
ec0mode=F100 was used (IRIX 6.3)
+o 482190: ftpd can be crashed, exposing clear text
password (IRIX 6.2, 6.3, 6.4)
+o 482476: Several bugs in DHCP server
+o 483646: panic in in_ifscrub() when hostname is "alias"
(IRIX 6.2, 6.3, 6.4)
+o 483963: large rcp/ftp transfers on O2 systems hang
(IRIX 6.3)
- 17 -
+o 484863: WIth -x duplicate names in ethers can be
created
+o 485482: Kernel panics with patch 1966 (IRIX 6.4)
+o 486283: IP fragmentation buggy under load
+o 487655: proclaim server dies with "file4 isn't open"
+o 487677: reproducible O2 kernel panic using xni
+o 487677: reproducible O2 kernel panic using xni (IRIX
6.3)
+o 489317: still possible to hang uninterruptably in
sosend()
+o 489360: tpisocket locking is still wrong
+o 490852: TLBMISS in m_free (IRIX 6.2, IRIX 6.4)
+o 493382: ftp ignores "ACCOUNT" in .netrc
+o 495807: NFS hangs with AIX client (IRIX 6.4)
+o 495864: ec2 driver problem (IRIX 6.2)
+o 497875: rcp I/O error after suspend (IRIX 6.4)
+o 498603: rlogin can coredump and reveal shadow passwords
(IRIX 6.2, 6.3, 6.4)
+o 498985: O2 with second ethernet interface (pci) (IRIX
6.3)
+o 499575: rlogin can core dump if TERM buffer is too long
(IRIX 6.2, 6.3, 6.4)
+o 499837: uipc_accept() could sleep in kmem_zone_alloc()
while locked (IRIX 6.4)
+o 500241: protocol failure in circuit setup (IRIX 6.2,
IRIX 6.4)
+o 500523: Cannot choose hostname/address when
hostname/address is default
+o 500939: AF_UNIX socket connect deadlock (IRIX 6.2, IRIX
6.4)
- 18 -
+o 501843: recursive mutex hang in DLPI code (IRIX 6.4)
+o 502760: portmap runs out of memory (IRIX 6.2, IRIX 6.3,
IRIX 6.4)
+o 504586: ypserv (6.3) dies/respawns often (IRIX 6.3)
+o 505540: kmem used after free - potential corruption
(IRIX 6.2, IRIX 6.4)
+o 506376: 6.2 rsvpd uses obsolete message format
+o 508398: rpcbind can dump core (IRIX 6.2, IRIX 6.3, IRIX
6.4)
+o 509380: 6.4 rsvpd uses obsolete message format (IRIX
6.4)
+o 512998: rpcbind stop servicing port 111
+o 513225: panic from in_pcbassign with patch 1703 (IRIX
6.2, IRIX 6.4)
+o 513595: rsvpd in 6.3 is obsolete (IRIX 6.4)
+o 514382: panic in mbscavenge() (IRIX 6.4)
+o 516917: Processes get stuck in SETSXBRK waiting on node
spefic kernel memory (IRIX 6.4)
+o 517472: mbuf usage goes up dramatically when patch 2019
installed (IRIX 6.4)
+o 519538: patched portmap complains PID table slot XX
empty
+o 520733: OCTANE/Onyx2 + patches panics in ioc3 driver
(IRIX 6.4)
+o 522016: bootp fails due to broadcast nonack
+o 523292: mace_ether_receive hang O2 (IRIX 6.3)
+o 525777: ftpd problems with '~' (IRIX 6.2, IRIX 6.3,
IRIX 6.4)
+o 526256: Octane crashes with BIT3 vme (IRIX 6.4)
+o 533815: DHCP assigns host name based on NIS domain
- 19 -
+o 536542: netstat CORE dump on 6.2 indy patchset#25 (IRIX
6.2, 6.3, 6.4)
+o 540871: Irix 6.2 ftp / ftpd not y2k compliant (IRIX
6.2, 6.3, 6.4)
+o 541451: rsvpd needs to be updated to rel4.1a6 for new
API (IRIX 6.2, 6.3, 6.4)
+o 542760: panic on bad vsocket (IRIX 6.4)
+o 545698: SPANS implementation for ATM-XIO card needed in
IRIX (IRIX 6.4)
+o 548001: Patch 2446 breaks Diskless clients on 6.3 (IRIX
6.3)
+o 548138: timeslave should know about a GPS receiver
(IRIX 6.2, IRIX 6.4)
+o 549465: multicast routing packets have bad checksum
(IRIX 6.2)
+o 552707: bad UDP checksums from EPLEX board (IRIX 6.2)
+o 554541: no support for Client ID and static allocation
of Ip addresses
+o 554888: DHCP front-end needs to support mdbm
+o 555856: timeslave broken with year=>2000 (IRIX 6.2,
6.3, 6.4)
+o 558302: `timeslave -Y` sets the year wrong (IRIX 6.2,
6.3, 6.4)
+o 559199: DHCP server assigns NetBIOS name server, but
misses node type
+o 559829: NFS copy corruption (block boundary
replacement) (IRIX 6.4)
+o 452978: dprof does not recognize counter-busy condition
+o 567877: _RLD_LIST order should be reversed for dplace,
dlook and dprof. Fixes interaction with other
environments such as MPI.
+o 582031: Rapid fork() calls result in poor memory
placement.
- 20 -
+o 588730: A bug in the IRIX 6.5 scheduler prevents
applications that use asynchronous IO and gang
scheduling to work together. To work around this
problem with applications that use libmp, set the
environment variable OMP_DYNAMIC to FALSE.
+o 589921: Panic in sysmp(MP_MUSTRUN_PID, ...) call
+o 598577: Panic in gang code while using miser
+o 603435: miser_cpuset core dumps
+o 603810: schedctl (NDPRI, ...) returns bogus previous
value.
+o 609302: Miser does not always start jobs on schedule.
+o 609303: Miser daemon incorrectly accounts information.
+o 705918: AN FSTAT OF C CREATED SOCKET RETURNS DIFFERENT
RESULTS (IRIX 6.3, 6.4)
4.4 _O_p_e_r_a_t_i_n_g__S_y_s_t_e_m
4.4.1 _B_u_g_s__f_i_x_e_d__i_n__I_R_I_X__6_._5_._2
+o 487889: removed the misleading "PRDA color mismatch"
warning message.
+o 597782: reintegrate code from 6.4 patch 2299 into 6.5.
Adds hwgraph vertices for big endian and little endian
devices on the pci bus.
+o 600578: While transferring data to adapters on the same
PCI bus as an AIC-1160 Fibre Channel adapter, it was
possible for the Fibre Channel adapter to mistake a
transaction as affecting it. This mistake would cause
two adapters to drive the bus at the same time,
producing PCI errors. The AIC-1160 are now addressed
in a different way, using I/O mapping instead of memory
mapping, in order to avoid this problem.
+o 614247: The dksc driver now returns requests with error
if they are too large for the extended read/write
commands. For a 512 byte sector disk, the maximum
transfer size is 33553920 bytes.
+o 617705: 3rd party disk drivers used in XLV volumes can
cause panic during attempted failover.
- 21 -
+o 620645: kernel heap allocator free routine got a NULL
Pointer
+o 623300: boot hangs or long boot times on large systems,
due to lock contention
+o 624579: hub error parsing could fail, interfering with
error recovery
+o 628367: error when "npri -w" is run on a process which
is already weightless
+o 632761: replace peer-to-peer dma capability for the O2
+o 583867, 635945: Three IPC semaphores parameters were
removed in IRIX 6.5, but the information in the book
"IRIX Admin: System Configuration and Operation" was
not revised until now. The parameters "semmns,"
"semume," and "summnu" have been deleted. "semume" and
"semmnu" are no longer valid because undo operations
are performed in a different way; there are no new
equivalents. "semmns" was a boot time table sizing
parameter, and these structures are now dynamically
allocated. The following IPC semaphores parameters
have changed default values: semmni (300), semmsl
(100), semopm (100).
+o 636641: The policy module create function (pm_create())
called the corresponding destroy functions if it failed
to create a policy module. None of the corresponding
destroy functions were set up to handle a partially
created policy module (i.e., they ended up referencing
null pointers). This fix remedies this situation.
4.4.2 _B_u_g_s__f_i_x_e_d__i_n__I_R_I_X__6_._5_._1
+o 235729: syslogd/syslod() loses priority info when
messages > 512 bytes
+o 338261: wtmp/utmp management make them unreliable
501830: all 'from' fields get zeroed in wtmpx w/ heavy
logins Numerous race conditions in libc's user database
implementation have been eliminated.
+o 545707: Correct hang on STK 4781/4791 when no media
present.
+o 558262: If an attempt (on exec(2)) to map in rld(1)
fails (perhaps because all file descriptors had been
used up before the exec) a coredump would be produced
in base irix6.5. Now an appropriate message is produced
- 22 -
and there is no coredump. The process returns a non-
zero status code (to wait(2) waitpid(2) or wait3(2),
for example) as the process status (because the
execution failed).
+o 561672: Add support for Sony GY-2120 which is the
replacement for the Sony Gy-10. Sony no longer markets
the GY-10.
+o 567126: Fix hang in Octane gfx code during NMI
+o 571933: sproc/tmpname fails
+o 578019: In a multibyte locale (MB_CUR_MAX != 1),
regexec(3G) would fill the pmatch array with the
position of parenthesized subexpressions as character
indices instead of byte indices. Now, it appropriately
returns the positions in bytes instead of characters.
+o 572255: Fix several bugs causing long delays when
booting large Origin 2000s.
+o 573430: C++-imcompatibility in curses.h
+o 582054: Allow more data to be sent thru the msgsnd()
system call.
+o 585514: Single-user mode login is no longer dependent
upon libdce.so when DCE integrated login binaries are
installed.
+o 588650: Fix some racy code to cure process hangs in
msgsnd()/msgrcv() calls.
+o 593333: The second port of a MSCSI card in a
GIGAchannel expansion would not negotiate wide data
transfer.
+o 594585: When cpu(s) were disabled on large Origin
2000s, the system would appear to hang while booting
for up to several minutes.
+o 594955: Added non-FIFO Centronics mode and EPP address
mode support to plp driver for O2.
+o 595731: Fix hang on Octane during NMI on 6.4 with
patches 1953 and 2895 installed.
+o 596948: Fix a kernel panic that could occur while
servicing a PIOCMAP procfs ioctl.
- 23 -
+o 597708: raid(1m) man page is deleted as no longer
supported.
+o 601104: CPR needs to operate with NQE + Miser session.
Add a hook to allow miser job to restart.
+o 606051: Preserve debugging information obtainable via
PIOCOPENM ioctl for preloaded objects across forks.
+o 606441: fixing regression with 6.4.1 2999 patch.
+o 606466: Bring Irix 6.5.1 up to the level of TDR6 for
Irix 6.4.1.
+o 607200: Make a special case for cluster
checkpoint/restart. If all the processes are within one
machine, no array function got involved.
+o 608092: Fixed file system corruption when passing
b_bcount > maxdmasz down to xlv or the disk
driver.
+o 608182: For miser jobs, account for pages shared
between processes in the same job without multiply
counting such pages.
+o 610581: panic with bad pcidma address
+o 610979: O2K-VME regressions from 6.4
+o 611000: O2K-VME DMA engine regressions from 6.4
4.5 _V_o_l_u_m_e__M_a_n_a_g_e_r
4.5.1 _B_u_g_s__f_i_x_e_d__i_n__I_R_I_X__6_._2__t_h_r_o_u_g_h__I_R_I_X__6_._5
+o 250074: Added a _c_r_e_a_t_e command to xlv_make(1m) for
writing newly defined objects out to affected disks
without exiting the tool.
+o 267257: Corrected misleading xlv_make(1m) error
message regarding missing block and/or character device
node.
+o 290849: XLV displays the nodename assoicated with a
volume.
+o 294565: XLV device nodes were created even if the
volume was incomplete, possibly leading to the mistaken
conclusion that the volume was usable.
- 24 -
+o 311156, 333694: Added a -_s option to xlv_assemble(1m)
so a list of logical volumes to assemble can be
specified on the command line. It is no longer
necessary to assemble all logical volumes at once.
This option assists failover procedures for the
FailSafe Product.
+o 312367: Added a -_h command line option to xlv_make(1m)
so a nodename can be specified. This option implies
the -_A option to not automatically invoke
xlv_assemble(1m).
+o 323922: Xlv_shutdown(1m) complains and fails with a
exit code of 1 when an invalid volume name is specified
with the -_n option.
+o 331525: Added three new commands to xlv_mgr(1m). The
_c_h_a_n_g_e _n_a_m_e command renames a xlv object. The _c_h_a_n_g_e
_t_y_p_e command changes the type of an object. The _c_h_a_n_g_e
_v_e__s_t_a_r_t command changes start address of a standalone
volume element.
+o 341023: Xlv_assemble(1m) won't gratuitously initiate a
plex revive for a triple-plexed volume element when the
volume element was cleanly shutdown.
+o 355237, 357554: The system can sometimes panic when a
subvolume goes offline.
+o 363121: Fixes a kernel panic if plexing software is not
installed and an I/O error occurs on a RAID device.
+o 370878: Fail over to an alternate controller on a dual
controller SGI RAID device only works on lun 0. Luns
other than 0 were not properly switched over to the
alternate path.
+o 375080: XLV can deadlock the system under high load
when it grabs locks which prevent I/O while running
configuration commands (i.e, xlv_assemble). The
xlv_assemble process may need then to acquire memory,
causing the system to need to do I/O, resulting in
deadlock.
+o 382630: In two different windows, simultaneously using
xlv_shutdown(1m) and xlv_mgr(1m) to remove a volume can
cause the a system panic.
+o 392195: Losing a disk drive that is part of a root
volume can sometimes cause the system to hang.
- 25 -
+o 405757: Fixed a problem in xlv_shutdown(1m) which
caused it to fail with out of memory errors on large
systems.
+o 422453: Changed xlv_assemble to exit with a non-zero
exit status if selective assemble can't assemble all of
the volumes specified on the command line.
+o 422766: Xlv_make(1m) failed to create a standalone
plex or volume element when the inter-mixing disk
partitions of types "xlv" and "xfs".
+o 422798: A plexed root drive could be corrupted if one
plex was detached/attached, and the system was then
cleanly shutdown and rebooted.
+o 426966: Xlv_mgr(1m) core dumps detaching a NULL volume
element from a standalone plex.
+o 438539: The XLV label daemon is pinned in memory and
allocates and frees a large structure (2MB). Since it
is pinned, the heap never shrinks, causing xlv_labd(1m)
to be a memory hog.
+o 471909: During system startup, if /tmp is a volume and
there are plexed volumes being revived, the mount of
the /tmp volume fails. /tmp cannot mount until all the
plexed volumes are revived.
+o 474748: During a XLV plex revive, a read error on the
source volume element causes the system to panic with a
kernel fault.
+o 528089: XLV doesn't process multiple plex errors
correctly when receiving many disk errors, and can
cause a kernel panic.
+o 532078: A root volume which is either concatenated or
striped isn't properly rejected at boot time.
+o 533693: xlv_mgr(1m) doesn't handle volumes with
duplicate names but different hostnames.
+o 534748: The death of the xlv_labd(1m) process can
cause the kernel to hang. We work around this by
restarting the deamon as necessary.
+o 548381: Incomplete volume elements can cause a kernel
panic. Occurs on mirrored plexes where different
portions of the address range are valid on the
different plex halves.
- 26 -
+o 569117: XLV utilities xlv_mgr(1m) and xlv_make(1m)
need to be compiled N32.
4.6 _X_F_S__F_i_l_e__S_y_s_t_e_m
4.6.1 _B_u_g_s__f_i_x_e_d__i_n__I_R_I_X__6_._5_._2
+o 622220 : Xfsrestore hangs in directory creation on
very large systems.
+o 626381 : Use of the '-t' option in xfsrestore could
result in removed files.
+o 626404 : Universal Font Manager (libufm.so) did not
provide vector outline approximations and exact
outlines at the same time.
4.6.2 _B_u_g_s__f_i_x_e_d__i_n__I_R_I_X__6_._5_._1
+o 502802: New tunable xfs_refcache_percent added to allow
the size of the NFS/XFS inode reference cache to be
adjusted (primarily for NFS fileservers supporting
quotas over NFS).
+o 536065: Xfsdump progress reports failing when dumping
to stdout.
+o 546301: Xfsrestore aborts after a bad read.
+o 564234: Dump inventory searches with xfsdump/xfsrestore
based on a mount point would quit after finding the
first match.
+o 566573: A process writing to a file using a buffer that
is memory mapped to that same file could deadlock.
+o 598085: Specfs can get into a race condition between
close and open of a craylink device.
+o 600228: A mechanism is needed to free up metadata
buffers in low memory situations.
+o 600394: ISO-9660 "Level 3" CD-ROM not readable.
+o 600716: Possible XFS panic under heavy system load when
file space has been reserved using the fcntl(F_RESVSP)
command and then the file is truncated and grown.
+o 603750: Heavy load on server creates processes that
can't be killed.
- 27 -
+o 604149: Creating a symlink in a full filesystem that is
also DMAPI-enabled could result in a crash.
+o 604483: Deadlock when doing I/O to the same memory
mapped file.
+o 614319: Xfs_repair core dumps on an unexpected inode
type.
+o 615536: Fcntl allocates wrong number of blocks in some
circumstances.
+o 616138: Xfs_rename looping in very active directory.
+o 617933: Update relnotes to document min_bufmem tunable
+o 536662: CL_BOTTOM_UP parameter used for impact
compression causes kernel panic.
+o 578236: On an Octane MXI with an IMPACT Compression
board decompressing JPEG image using the hardware
introduces 2 rows of black pixels at the bottom of the
image.