Technical article 125084

My kernel won't relink after an Openserver 5 in-place upgrade

OpenServer 5.0.5OpenServer 5.0.6

Written · Last revised

Problem

After performing an in-place upgrade from 5.0.5 to 5.0.6, the kernel relink fails with

"/var/opt/K/SCO/link/1.1.1Ga/conf/pack.d/clock/space.c", line 356:
 warning: undeclared initializer name CORE_CREATE_REPLACE
"/var/opt/K/SCO/link/1.1.1Ga/conf/pack.d/clock/space.c", line 356:
 warning: illegal initialization
"/var/opt/K/SCO/link/1.1.1Ga/conf/pack.d/clock/space.c" warning: will
 not compile properly

i386ld Driver.o fatal: can't open file /var/opt/K/SCO/link/1.1.1Ga/conf/pack.d/clock/space.o for input

Error:  can not link-edit unix

Cause

A likely cause is that the in-place upgrade did not succeed due to corruption in the software database (custom). In this case there was a failure to fully upgrade the Unix component due to one of the Unix sub-components being marked as corrupt. As such, the file /usr/include/sys/signal.h was still pointing to the 5.0.5 signal.h instead of pointing to the new 5.0.6 signal.h which defines CORE_CREATE_REPLACE.

Examination of the custom database prior to the upgrade showed that the Unix SharedControl package was marked as corrupt.

From /var/opt/K/SCO/Unix/5.0.5Eb/.softmgmt/client.db :

DBASEINFO: type=CLIENT-COMPONENT version=1.0 COMP:SCO:Unix: version=5.0.5Eb installStd=SSO-1 patches=rs505a.Unix505.1.0a, RS505A: Core OS Supplement, oss497a.Unix505.1.0a, OSS497A: Base OS Supplement, oss497b.Unix505.1.0a, OSS497B: Base OS Supplement, oss600a.Unix505, Y2K Patch to RS505A for Unix, oss497c.Unix505.1.0a, OSS497C: Base OS Supplement, OSS642a.506.SCO.Unix.RTS, OSS642a for 506\, 505\, 504 description=SCO UNIX System V Operating System state=PART-ATTACHED, PART-CORRUPT-ATTACHED, PART-REGISTERED, PART-CORRUPT-REGISTERED, PART-CONFIGURED, PART-CORRUPT-CONFIGURED numFiles=1336 subpackages=SCO:Unix:SharedControl, SCO:Unix:ClientControl, SCO:Unix:BASEOS, SCO:Unix:EXTUTILS, SCO:Unix:SYSADM, SCO:Unix:SOFT, SCO:Unix:INTERNET, SCO:Unix:OPTIONALS, SCO:Unix:OS_DOC required=SCO:Unix:SharedControl, SCO:Unix:ClientControl, SCO:Unix:BASEOS PKG:SCO:Unix:SharedControl: description=Installation directory state=FULL-ATTACHED, CORRUPT-ATTACHED, FULL-REGISTERED, CORRUPT-REGISTERED, FULL-CONFIGURED, CORRUPT-CONFIGURED affectingPatches=rs505a.Unix505.1.0a, oss497a.Unix505.1.0a, oss497b.Unix505.1.0a, oss600a.Unix505, oss497c.Unix505.1.0a, OSS642a.506.SCO.Unix.RTS ... ... ...

You can see in the "state=" lines where the PKG:SCO:Unix:SharedControl: reports

state=FULL-ATTACHED, CORRUPT-ATTACHED, FULL-REGISTERED, CORRUPT-REGISTERED,
FULL-CONFIGURED,CORRUPT-CONFIGURED
Normally, this should read
state=FULL-ATTACHED,  FULL-REGISTERED,  FULL-CONFIGURED

And, since this is marked as corrupted, the parent component, COMP:SCO:Unix:, is also marked partially corrupted:

state=PART-ATTACHED, PART-CORRUPT-ATTACHED, PART-REGISTERED,
PART-CORRUPT-REGISTERED, PART-CONFIGURED, PART-CORRUPT-CONFIGURED
Normally, this should read
state=PART-ATTACHED,  PART-REGISTERED,  PART-CONFIGURED

Solution

To determine if there is any corruption in the custom database, you can run the following commands

# cd /var/opt/K/SCO
# find . -follow -type f -name "*.db" | xargs fgrep -l CORRUPT
and
# cd /opt/K/SCO
# find . -follow -type f -name "*.db" | xargs fgrep -l CORRUPT

If either of these find commands return a file name, those file(s) will need to be examined for what the *CORRUPT* entries look like.

It is possible that components/packages may be marked as *CORRUPT* when there is nothing really wrong with them.If so, it may be possible to edit the database files and remove the *CORRUPT* portion of the entry as shown above. Setting these to non-CORRUPT states should allow the in-place upgrade to fully upgrade these packages/components.

****IMPORTANT NOTE****

Careful examination of the system should be done prior to attempting this type of procedure. These entries may be marked corrupt for a very valid reason and doing this may just mask problems that will occur later. A good start in checking the sanity of the system is is to run a Verify System -> Normal system state thorough from the software manager. Additionally, you may also want to manually check the files within that component/package. And, as always, make sure that you have good backups of the system in case there are problems with the database editing or the upgrade.

From the Xinuos technical article database, record 125084. Reproduced with permission. The original is at www.sco.com/ta/125084, which is reachable but blocked from search engines.

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.