On Wed, 3 Apr 2013 16:34:45 -0700, Christoffer Dall
<cdall@xxxxxxxxxxxxxxx>
wrote:
> there was a similar problem with the kvm 32-bit makefile at some
> point, the fix looks ok to me.
>
> -Christoffer
>
> On Wed, Apr 3, 2013 at 4:28 PM, Geoff Levand <geoff@xxxxxxxxxxxxx>
wrote:
>> Hi Michal,
>>
>> I was building Marc's kvm development kernel outside of the
>> sources (make O=), and I got the following build error:
>>
>> arch/arm64/kvm/../../../arch/arm/kvm/mmio.c: fatal error: opening
>> dependency file arch/arm64/kvm/../../../arch/arm/kvm/.mmio.o.d: No
such
>> file or directory
>>
>> The cause was that kbuild did not create the output directory
>> arch/arm64/kvm/../../../arch/arm/kvm, and so gcc failed.
>>
>> Here is what was in his makefile:
>>
>> obj-$(CONFIG_KVM_ARM_HOST) += $(addprefix ../../../virt/kvm/,
>> kvm_main.o coalesced_mmio.o)
>>
>> I changed this to build a composite and kbuild then created
>> the output directories:
>>
>> obj-$(CONFIG_KVM_ARM_HOST) += kvm.o
>> kvm-$(CONFIG_KVM_ARM_HOST) += $(addprefix ../../../virt/kvm/,
>> kvm_main.o coalesced_mmio.o)
>>
>> I guess kbuild used some other rules and saw the dependency on
>> the output directory.
>>
>> So, my questions are:
>>
>> Is it expected that the original rule should fail?
>> Is my fix the correct way?
>> Can I use some other way to get those output directories created?
Yeah, this looks like the thing to do. I'll fix the KVM/arm64 makefile.
Thanks for investigating.
M.
--
Fast, cheap, reliable. Pick two.
_______________________________________________
kvmarm mailing list
kvmarm@xxxxxxxxxxxxxxxxxxxxx
https://lists.cs.columbia.edu/cucslists/listinfo/kvmarm
