cvs commit: src/sys/i386/include param.h

[Paul Seniura]

Hi,


Building modules-with-world has been broken by these
commits (v1.74 & v1.75 of src/sys/i386/include/param.h). 
Scott's v1.75 patch is insufficient to fix this. 
Here's a snip of my log with v1.75:


[...]
===> 3dfx
rm -f .depend
mkdep -f .depend -a   -nostdinc -D_KERNEL -DKLD_MODULE -I- -I. -I@ -I@/contrib/altq -I@/../include /src/sys/modules/3dfx/../../dev/tdfx/tdfx_pci.c
In file included from @/sys/param.h:109,
                 from /src/sys/modules/3dfx/../../dev/tdfx/tdfx_pci.c:42:
./machine/param.h:102:23: opt_sched.h: No such file or directory
mkdep: compile failed
*** Error code 1 (continuing)
`depend' not remade because of errors.
===> aac
rm -f .depend
mkdep -f .depend -a   -nostdinc -D_KERNEL -DKLD_MODULE -I- -I. -I@ -I@/contrib/altq -I@/../include /src/sys/modules/aac/../../dev/aac/aac.c /src/sys/modules/aac/../../dev/aac/aac_pci.c /src/sys/modules/aac/../../dev/aac/aac_disk.c /src/sys/modules/aac/../../dev/aac/aac_cam.c
In file included from @/sys/param.h:109,
                 from /src/sys/modules/aac/../../dev/aac/aac.c:40:
./machine/param.h:102:23: opt_sched.h: No such file or directory
In file included from @/sys/param.h:109,
                 from /src/sys/modules/aac/../../dev/aac/aac_pci.c:39:
./machine/param.h:102:23: opt_sched.h: No such file or directory
In file included from @/sys/param.h:109,
                 from /src/sys/modules/aac/../../dev/aac/aac_disk.c:35:
./machine/param.h:102:23: opt_sched.h: No such file or directory
In file included from @/sys/param.h:109,
                 from /src/sys/modules/aac/../../dev/aac/aac_cam.c:36:
./machine/param.h:102:23: opt_sched.h: No such file or directory
mkdep: compile failed
*** Error code 1 (continuing)
`depend' not remade because of errors.
===> accf_data
rm -f .depend
mkdep -f .depend -a   -nostdinc -D_KERNEL -DKLD_MODULE -I- -I. -I@ -I@/contrib/altq -I@/../include /src/sys/modules/accf_data/../../netinet/accf_data.c
In file included from @/sys/param.h:109,
                 from /src/sys/modules/accf_data/../../netinet/accf_data.c:31:
./machine/param.h:102:23: opt_sched.h: No such file or directory
mkdep: compile failed
*** Error code 1 (continuing)
`depend' not remade because of errors.
===> accf_http
[...]


I'm able to build a kernel with v1.75 just fine (I'm about
to do that and test it RSN :) ).

Going back to v1.73 or v1.72 of src/sys/i386/include/param.h
_will_ allow building modules-with-world.  v1.73 of course
has #define PREEMPTION with no wrapping logic at all, and
v1.72 does not define it at all period.  v1.73 seems to
therefore define it such that other modules can pick it up
when building modules-with-world, and would be more in-sync
with the kernel build this way.

My World is current as of yesterday's CTM deltas; since the
deltas are 16-hours apart, no time for today's delta to be
built & installed -- and now that's not going to happen
until src/sys/i386/include/param.h is fixed properly.  ;)

I've gone on record here on this list requesting the
official tinderboxes attempt building modules-with-world at
least once every run and pass the reports up the usual way. 
The puny p2 box I'm using seems to be the only tinderbox
doing "modules-with-world".  Hint hint.  ;)

Thank you & everyone for the work going into PREEMPTION
and everything else.


  --  thx, Paul Seniura
           System Specialist
           State of Okla. D.O.T.