Technical article 109975

What is SLS OSS469D, the Core OS Supplement for SCO OpenServer Release 5.0.4?

OpenServer 5.0.4

Written · Last revised

Problem

What problems are addressed by Support Level Supplement (SLS) OSS469D, the Core OS Supplement for SCO OpenServer Release 5.0.4?

Solution

SLS OSS469D addresses several known issues with the kernel, networking, and SMP components of SCO OpenServer Release 5.0.4 operating systems.

See the attached cover letter for more information about this SLS, including a list of specific problems addressed and installation instructions.

Dear SCO Customer,

Support Level Supplement (SLS) OSS469D, the Core OS and Networking Supplement, corrects problems in core operating system functionality and in networking. See the list below for more details on specific problems addressed by this SLS.

SLS OSS469D is intended for use on these products:

SCO OpenServer Enterprise System Release 5.0.4 SCO OpenServer Desktop System Release 5.0.4 SCO OpenServer Host System Release 5.0.4

Note: To obtain some of the same corrections for SCO OpenServer 5.0.0, 5.0.2 and Internet FastStart 1.0.0, 1.1.0, see SLS OSS468A. To obtain some of the same corrections for SCO OpenServer 5.0.5, see SLS OSS497C.

On SCO OpenServer 5.0.4, the following supplement MUST be installed prior to installing SLS OSS469D:

Release Supplement 504C for SCO OpenServer Release 5.0.4 (rs504c) http://www.caldea.com/support/download.html

It is not necessary to remove earlier versions of OSS469 prior to installing OSS469D.

SLS OSS469D supersedes the SCO Security Patch SSE010, a patch for IP denial of service vulnerabilities.

SLS OSS469D supersedes OSS485A, the Memory Supplement for OpenServer 5.0.4.

SLS OSS469D corrects the following problems, including all problems corrected in prior releases of OSS469:

Note: OSS469A corrected the following problems:

1. System "hang" (no keyboard response or other apparent system activity) under a network load which may be followed, after several minutes, by a warning message similar to:

WARNING: IP: spinning on pcb ...
This only occurs on multi-processor systems, and may be accompanied
by a depletion of streams buffers.  SCO Tracking number: 710315
2. Lock timeout panic in "narrowcast" or "cpuintr"; this only occurs on
   multi-processor systems.  SCO Tracking number: 710315

3. Unexpected Out of Band Data denial-of-service attack; an exploit has been distributed under the name "winnuke".

4. Spoofed Self-connect Packets denial-of-service attack; exploits have been distributed under the names "land" and "latierra".

Note: SCO OpenServer systems are NOT vulnerable to IP denial-of-service
attacks based on fragmented packets with invalid sizes.  Exploits have
been distributed under the names "teardrop", "bonk", and "newtear".

5. Warning message "WARNING: NLM: request to server failed:", followed by a panic in "printf" called from the NFS lock manager. SCO Tracking number: 500585

6. Incoming udp broadcasts can bring down a system by using up all its
   streams buffers.  This may be occurring because a program, such as
   scologin, has stopped reading incoming packets. Or it may be caused
   by a flood of incoming packets that are arriving too fast for a slow
   system.  SCO Tracking number: 710391
7. AFPS performance problem. Setting the option TCP_NODELAY improves
   performance of certain size transactions, which would otherwise show
   large degradation in performance.  SCO Tracking number: 500495

8. UNIX domain sockets performance problem. Performance of UNIX domain sockets would degrade with multiple CPUs compared with a single CPU. SCO Tracking number: 710402

Note: OSS469B corrected these additional problems:
9. TCP performance in certain environments.  The symptom of this problem
   is large numbers of retransmit timeouts, as shown by "netstat -sptcp."
   SCO Tracking number: 500604

You will need to edit /etc/conf/pack.d/tcp/space.c to enable the new behavior. Increase the parameter "tcp_fastretx_max" in increments of 3 and observe if this reduces the number of retransmit timeouts and improves performance.

10. Corrects a problem with terminal output disabling interrupts for excessive periods of time, which could lead to the clock losing time, or to warning messages such as, "too many sptfrees while flushing TLBs". SCO Tracking number: 710468

11. Corrects a panic in the polltime() routine. SCO Tracking number: 500570

12. Fixes a lock timeout panic in doflush(). SCO Tracking number: 500570

13. Corrects problems in the apic driver with sharing of interrupts.
    Fixes problems in which the system would not complete its startup
    sequence and would hang at the letter 'G'.  SCO Tracking
    number: 710566

14. Corrects a panic in "iknt_to_stream()". SCO Tracking number: 500732

15. Corrects a problem in the initialization of the apic hardware
    which resulted in a panic with the message: "Not all CPUs have
    initialized their interrupt controllers".
    SCO Tracking numbers: 710566

16. Corrects a lock timeout in sptfree().

Note: OSS469C corrected these additional problems:
17. Corrects a panic in kmem_alloc(), poll(), strpoll(), or clockb()
    called from queuerun().  SCO Tracking number: 710657
18. Corrects a panic in tcp_linput(), or in_pcballoc() called from
    tcp_linput().  SCO Tracking number: 710699

19. Corrects a failure to boot, following installation of OSS469B, with the following possible symptoms:

- "WARNING: CPU2 boot failure" - Hang at "E mp_setup" on startup

SCO Tracking number: 500871

20. Eliminates the following warning message on startup:

WARNING: differing CPUs, can't chain interrupts together
SCO Tracking number: 500871
21. Corrects a failure to allocate sufficient memory for the "streams"
    subsystem with values of NSTRPAGES greater than 4000.  The limit
    of NSTRPAGES is now 8000.  SCO Tracking number: 710502

22. Corrects a problem which can lead to the following warning message being displayed:

too many sptfrees while flushing TLBs

Note: This warning message can occur for different reasons. Some systems will continue to display this message following installation of OSS469C. If the message does not occur repeatedly, it can be safely ignored. SCO Tracking numbers: 710712

23. Corrects a problem in which networking programs using the TLI
    interface fail to run.  SCO Tracking number: 710718
24. Corrects a problem where the various <Lock> states (Caps, Num, Scroll)
    could disagree with the keyboard lights after changing window focus
    on a graphical screen.
    SCO Tracking numbers: 710525
25. Corrects a problem in which deleting circular routes would cause a
    kernel panic.  SCO Tracking number: 500854
Note: OSS469D corrects these additional problems:
26. 'brand -g -a' coredumped if the nodelock file was empty, instead
    of populating it.  SCO Tracking numbers: SCO-236-1427, SCO-236-1421,
    710673
27. If the login delay (the t_logdelay field in /etc/auth/system/default)
    was set to zero, the passwd generator would generate passwords of an
    incorrect length.  SCO Tracking numbers: SCO-236-904, 710643

28. The cron utility would get "stuck" waiting for long jobs to complete, blocking its ability to accept and run new "at now" jobs promptly. SCO Tracking numbers: LTD-246-1342, 500904

29. SCO PPP with dynamic CHAP authentication would fail.  Wildcard '*'
    can now be used.  SCO Tracking numbers: SCO-236-1167, 710427
30. Running a large number of simultaneous fsck's after an irregular
    shutdown could result in a kernel panic.
    SCO Tracking numbers: SCO-236-1397, 500802

31. hpnptyd ran in an endless loop. SCO Tracking numbers: SCO-247-112, 500794

32. The Bourne shell(sh) improperly handled window size changes. SCO Tracking number: SCO-59-9705

33. Certain TLI and XTI connections hung during connection startup. SCO Tracking numbers: SCO-236-1447, LTD-246-1321, 710577, 710718

34. HTFS uses a "synchronous log" file named .slog000 to enhance the
    performance of synchronous writes.  This file was not intended to be
    visible to applications or users, but in some circumstances, could
    become visible, causing backup applications to fail.
    SCO Tracking numbers: SCO-236-1446, 500943
35. On multiprocessor systems, assigning an interrupt to a CPU could fail
    with the message, "iattr_add_to_shared_chain failed."
    SCO Tracking numbers: SCO-236-1469, 710769

36. Allows disabling of the warning message, "Too many sptfrees while flushing TLBs." SCO Tracking numbers: SCO-59-9912, 710820

Following installation of SLS OSS469D, the warning messages will
be disabled, by default.  The warnings are being disabled because
the condition indicated is not harmful.  If you want to re-enable
the warnings, you will need to edit /etc/conf/pack.d/kernel/space.c.
Change the parameter "suppress_sptfree_msg" from 1 to 0 to enable
warnings.
37. When running low on streams memory, all networking activity could
    cease for several minutes.  SCO Tracking numbers: SCO-59-9910, 710760

38. A panic can occur in "table_grow" when total memory is very limited. SCO Tracking number: SCO-59-9907

39. The following warning message could occur with certain serial I/O hardware: "kmem_alloc: called at interrupt time without KM_NOSLEEP." SCO Tracking number: SCO-59-9909

40. TCP applications in which the receiver had stopped reading could result
    in a depletion of streams memory.  SCO Tracking numbers: SCO-236-1467,
    710760
41. A panic could occur in "dlpi_unlink" under conditions in which the
    network card had stopped responding.
    SCO Tracking numbers: SCO-242-144, 710634

42. The link kit file "arp/space.c" contained the wrong definition of the data structure "arp_pcb[]", which might lead to a panic.

43. Allows "hpnpf" to talk on alternate ports.  It now supports "-p"
    port arguments 9100 through 9104.  SCO Tracking number: LTD-245-2801

44. login(C) and su(C) binaries fail to work properly with large group files. SCO Tracking number: 710666

45. A panic could occur in "badsleep" called from ttywait when using "ldterm". SCO Tracking numbers: LTD-245-2846, 501132

46. A panic could occur in "emunmap" on multi-processor systems. SCO Tracking numbers: LTD-246-1407, 501048

47. Certain parts of the kernel were unnecessarily insisting on allocating
    memory with addresses under 16Mb, which could exhaust this memory and
    cause poor system performance.
    SCO Tracking numbers: LTD-246-1302, 500722
SLS OSS469D contains these files:

/etc/brand /etc/ifor_pmd /etc/ifor_sld /etc/sco_cpd /bin/login /bin/su /bin/x286emul /bin/i286emul /bin/passwd /bin/sh /etc/cron /usr/lib/hpnp/hpnptyd /usr/lib/hpnp/hpnpf /etc/conf/pack.d/str/Driver.o /etc/conf/pack.d/str/space.c /etc/conf/pack.d/kernel/io.a /etc/conf/pack.d/kernel/os.a (sys2.o, valloc.o, page.o, startup.o, unistubs.o, kma.o, dynamic.o, sys3.o) /etc/conf/pack.d/kernel/space.c /etc/conf/pack.d/hpps/Driver.o /etc/conf/pack.d/sp/Driver.o /etc/conf/pack.d/bio/Driver.o /etc/conf/pack.d/ht/Driver.o /etc/conf/pack.d/ptem/Driver.o /etc/conf/pack.d/spt/Driver.o /etc/conf/pack.d/ht/space.c /etc/conf/pack.d/apic/Driver.o /etc/conf/pack.d/apic/space.c /etc/conf/pack.d/crllry/Driver.a (kentry.o, mppic.o, mpswtch.o, misc.o) /etc/conf/pack.d/mpsw/Driver.o /etc/conf/pack.d/Sdsk/Driver.o /etc/conf/pack.d/tcp/Driver.o /etc/conf/pack.d/tcp/space.c /etc/conf/pack.d/udp/Driver.o /etc/conf/pack.d/ip/Driver.o /etc/conf/pack.d/iknt/Driver.o /etc/conf/pack.d/rte/Driver.o /etc/conf/pack.d/Sdsk/Driver.o /etc/conf/pack.d/lockd/Driver.o /etc/conf/pack.d/net?/Driver.o /etc/conf/pack.d/arp/space.c /usr/lib/ppp/ppp /usr/bin/netstat /usr/include/sys/netinet/tcp_var.h /etc/nbd /etc/snmpd /etc/pppstat /etc/pppattach /etc/pppd

I. INSTALLING SLS OSS469D
1. Prepare the media for installation.
Download the OSS469D VOL files from the SLS/oss469d directory on SCO's
ftp site.  Place the files in the /tmp directory on the local server.
The files should include:
VOL.000.000
VOL.001.000
oss469d.ltr

2. Run the Software Manager with the command:

# scoadmin software

or double-click on the Software Manager icon in the desktop.

3. Pull down the "Software" menu and select "Install New".

4. When prompted for which host to install from, choose the local machine and then "Continue".

5. When prompted to select the media device:

If you are installing this SLS from diskette media, select the appropriate floppy disk drive and then "Continue". Proceed to step 6.

If you are installing this SLS from media images per step 1, select
"Media Images", then "Continue".  When prompted for the "Image
Directory", enter "/tmp" (or the directory where you placed the VOL
file in step 1) and then select "OK".

6. When prompted to select software to install, ensure that the "OSS469D" entry is highlighted. Choose "Install".

Note: If OSS469A or OSS469B was previously installed on your system,
you may see a warning during this phase of the installation about
an incorrect checksum on file ".../pack.d/kernel/io.a".  This warning
can be safely ignored.

7. You will be asked whether or not you want to relink the kernel.

Enter "y" or "n" based on this message. Note that you will not be able to make use of the patch until you relink the kernel. Therefore, it is suggested that you answer "y" to this question.

8. Once the installation finishes, quit the Software Manager.  Installation
   of SLS OSS469D is now complete. The changes will take effect the next
   time you reboot the system.

9. Log in as root on tty01, and shut down the system:

# shutdown -g0 -y

After shutting down and rebooting the machine, application of the SLS is complete.

II. REMOVING SLS OSS469D

Notes: Patches must be rolled back in the reverse order in which they were installed on a per-component basis.

After removing SLS OSS469D, the Software Manager may produce a "verification error" due to an unexpected checksum on file "io.a". This warning can be safely ignored.

1. Log in as root.
2. Execute the command:
# scoadmin software

or double-click on the Software Manager icon in the desktop.

3. Highlight "OSS469D: The Core OS Supplement".

4. Pull down the "Software" menu and select "Remove Software".

5. You will be asked whether or not you want to relink the kernel.

Enter "y" or "n" based on this message. Note that package removal
will not take effect until you relink the kernel.  Therefore, it
is suggested that you answer "y" to this question.

6. Once the removal finishes, quit the Software Manager.

7. As root on tty01, shut down the system:

# shutdown -g0 -y

After shutting down and rebooting the machine, removal of the SLS is complete.

If you have questions regarding this SLS, or the product on which it is installed, please contact your software supplier.

From the Xinuos technical article database, record 109975. Reproduced with permission. The original is at www.sco.com/ta/109975, which is reachable but blocked from search engines. Xinuos marks this article against a retired product line. That refers to the product’s lifecycle, not to whether the advice still works — plenty of it still applies to running systems.

Next step

Need this fixed rather than explained?

Tell us the release, what the system runs and what it is doing. We will reply by email with what we would do next.

  • You deal with the people who do the work
  • No obligation, and no upgrade required before we will help
  • We tell you plainly if something is not worth doing

Send an enquiry

We reply by email. If a server is down, say so in the first line. You can also write to help@scoexpert.com.