mk_kernel errors.

[rleon]

When I try to mk_kernel i get the following errors.


# /usr/sbin/mk_kernel -s system
WARNING: No such device, tunable, or cdio ehci.
Ignoring the following line.
ehci
WARNING: No such device, tunable, or cdio side_ba.
Ignoring the following line.
side_ba
WARNING: No such device, tunable, or cdio fcd_fcp.
Ignoring the following line.
fcd_fcp
WARNING: No such device, tunable, or cdio fcd_vbus.
Ignoring the following line.
fcd_vbus
Generating module: krm...
Generating module: gvid_info...
Generating module: drmfgl...
Generating module: gvid_him_fglrx...
Generating module: drmfglrx...
Compiling conf.c...
(Bundled) cc: "/usr/conf/space.h.d/system_space.h", line 564: error 1000: Unexpected symbol: "alwaysdump_tune".
(Bundled) cc: error 2017: Cannot recover from earlier errors, terminating.
*** Error exit code 1

Stop.
config: make did an exit(1)


Anyone have any ideas?

[Steven E. Protter]

Shalom,

Looks like a trashed system file and or devices being compiled into the kernel that do not exist.

May be missing files or incorrect driver configuration.

SEP

[skt_skt]

Looks like you have a kind of possible typying error.

here is what i see in my server

kmtune|grep -i dump
alwaysdump 1 - 1
dontdump 0 - 0

Which is os version

[rleon]

I have found out what the issue was. I took our test server) and ran

find / -name core*

I from that list I found some kernel related core files.
/usr/conf/master.d/core-hpux
/usr/conf/master.d/core_setid
/usr/conf/space.h.d/core-hpux.h
/usr/conf/sys/core.h
/usr/include/sys/core.h
/usr/include/core.h

I moved the files a tmp directory that I created and was able to duplicate the problem.
Then I only moved /usr/conf/master.d/core-hpux back and once again I was able to duplicated the problem to the point where we are now. Finally I moved the rest of the files back and everything worked fine.

Now on the system in question I confirmed that the files listed above where deleted. After coping the needed files over from a system at the same patch level I was able to successfully build a new kernel.