Am 08/22/12 17:24, schrieb Garrett Cooper:
> On Wed, Aug 22, 2012 at 7:02 AM, Chris Rees <[email protected]> wrote:
>> On 22 August 2012 14:32, O. Hartmann <[email protected]> 
>> wrote:
>>> On 08/22/12 15:05, Chris Rees wrote:
>>>> On 22 August 2012 08:18, O. Hartmann <[email protected]> wrote:
>>>>> Am 08/21/12 22:53, schrieb Bernhard Fröhlich:
>>>>>> On Di., 21. Aug. 2012 13:56:02 CEST, O. Hartmann 
>>>>>> <[email protected]> wrote:
>>>>>>
>>>>>>> Today I ran into a problem with FreeBSD 10.0-CURRENT #0 r239477
>>>>>>> recompiling the VBox kernel module:
>>>>>>>
>>>>>>> ===>   Cleaning for virtualbox-ose-kmod-4.1.18_1
>>>>>>> ===>   virtualbox-ose-kmod-4.1.18_1 is marked as broken: Does not 
>>>>>>> compile
>>>>>>> on FreeBSD < 8.2.
>>>>>>> *** [all] Error code 1
>>>>>>>
>>>>>>> Stop in /usr/ports/emulators/virtualbox-ose-kmod.
>>>>>>>
>>>>>>> The port compiled prior to the updates of today.
>>>>>>>
>>>>>>> oh
>>>>>>
>>>>>> What is the question? We have deprecated < 8.2 almost five months ago 
>>>>>> because of the memobj r0 rewrite. You're the first who complains about 
>>>>>> this so I'm pretty sure it was the correct decision to move forward. 
>>>>>> Please also keep in mind that 8.1 is already EOL.
>>>>>>
>>>>>
>>>>> I do not understand your complains about my mail to the list. For your
>>>>> information: I'm on FreeBSD 10.0-CURRENT and this seems, as far as I
>>>>> know, considered not to be deprecated by now. If I'm wrong in that, sorry.
>>>>
>>>> Can you give us the output of make -C
>>>> /usr/ports/emulators/virtualbox-ose-kmod -VOSVERSION please?
>>>
>>> 1000017
>>>
>>>>
>>>> I think your version of sys/param.h could be out of date.
>>>>
>>>> Chris
>>>
>>>
>>>
>>> As soon as I'm sitting in front of the machine again. But I doubt that
>>> the sys/param.h is out of date, since that box has been recompiling
>>> FreeBSD 10.0-CURRENT now for almost every day.
>>
>> Hm, does that mean that the output you gave me earlier was not from
>> the same machine?
>>
>>> When the problem occured, the change to CLANG 3.2-Beta has just been
>>> introduced, installed with "make buildworld installkernel". This
>>> procedere also triggers (via /etc/src.conf)
>>>
>>> PORTS_MODULES=          "x11/nvidia-driver"
>>> PORTS_MODULES+=         "emulators/virtualbox-ose-kmod"
>>
>> Please remove the quotes! [1]
>>
>>> the build of the modules for the nvidia GPU and Vbox kernel module.
>>>
>>> Thanks for your response,
>>
>> Something really weird is happening here.  All that is happening is
>> that the Makefile is testing against OSVERSION;
>>
>> .if ${OSVERSION} < 802000
>> BROKEN=         Does not compile on FreeBSD < 8.2
>> .endif
>>
>> Since the result from make -VOSVERSION shows that your OSVERSION is
>> indeed above 802000, I am completely at a loss as to what's going on
>> :(
> 
> I think this is the issue that I was referring to earlier that I noted
> to dougb@ but hadn't gotten an opportunity to repro because I've been
> busy with other things.
> 
> What does OSVERSION look like when you run it from make buildkernel
> (use this patch)?
> 
> Thanks,
> -Garrett
> 
> $ svn diff /usr/src/sys/conf/kern.post.mk
> Index: /usr/src/sys/conf/kern.post.mk
> ===================================================================
> --- /usr/src/sys/conf/kern.post.mk      (revision 239292)
> +++ /usr/src/sys/conf/kern.post.mk      (working copy)
> @@ -61,6 +61,8 @@
>         OSVERSION=${OSRELDATE} \
>         WRKDIRPREFIX=${WRKDIRPREFIX}
> 
> +.warning "PORTSMODULESENV => ${PORTSMODULESENV}"
> +
>  # The WRKDIR needs to be cleaned before building, and trying to
> change the target
>  # with a :C pattern below results in install -> instclean
>  all:
> 

On both systems I have to maintain, the output is:

BOX 1, patched:
make -C /usr/ports/emulators/virtualbox-ose-kmod -VOSVERSION
1000017

BOX 2, unpatched:
make -C /usr/ports/emulators/virtualbox-ose-kmod -VOSVERSION
1000017

No difference.

As I reported Chris Rees a moment ago, the "faulty" box was running a
kernel compiled with CLANG 3.1 while I installed world with the freshly
compiled CLANG 3.2 stuff and then tried to rebuild the kernel module for
VBox with this mixture. The CLANG issue is a wild guess. I didn't follow
the install procedure for a new world (single user mode startup with new
kernel, installworld, mergemaster et cetera). I guess something really
nasty happened that way and is possibly not worth to be investigated
further.

After I rebuild and installed world on my lab's box today the regular
way and faced no problems building the kernel module for virtual box, I
guess the problem is none but chasing ghosts ...

Regards,
Oliver


Attachment: signature.asc
Description: OpenPGP digital signature

Reply via email to