Build error for nommu if CONFIG_DEBUG_ATOMIC_SLEEP is selected

[Michael S. Tsirkin]

On Fri, Oct 18, 2013 at 10:21:16PM +0800, Axel Lin wrote:
> 2013/10/18 Michael S. Tsirkin <mst at redhat.com>:
> > On Fri, Oct 18, 2013 at 05:27:55PM +0800, Axel Lin wrote:
> >> Hi Michael,
> >>
> >> I got below build error for nommu if CONFIG_DEBUG_ATOMIC_SLEEP is selected.
> >> Also got the same build error if CONFIG_PROVE_LOCKING is selected.
> >>
> >> Seems this issue is introduced by commit 662bbcb2747c2
> >> "mm, sched: Allow uaccess in atomic with pagefault_disable()".
> >>
> >> arch/arm/kernel/built-in.o: In function `arch_ptrace':
> >> arch/arm/kernel/ptrace.c:852: undefined reference to `might_fault'
> >> arch/arm/kernel/built-in.o: In function `restore_sigframe':
> >> arch/arm/kernel/signal.c:173: undefined reference to `might_fault'
> >> arch/arm/kernel/signal.c:174: undefined reference to `might_fault'
> >> arch/arm/kernel/signal.c:175: undefined reference to `might_fault'
> >> arch/arm/kernel/signal.c:176: undefined reference to `might_fault'
> >> arch/arm/kernel/built-in.o:arch/arm/kernel/signal.c:177: more undefined references to `might_fault' follow
> >> make: *** [vmlinux] Error 1
> >>
> >> Regards,
> >> Axel
> >
> > Can you send/upload your full .config please?
> 
> I patch the kernel to port it for an arm7tdmi SoC.
> But I think you can easily reproduce it by simply
> select ARM7TDMI cpu and then enable either
> CONFIG_DEBUG_ATOMIC_SLEEP or CONFIG_PROVE_LOCKING.
> 
> Note, might_fault is implemented in mm/memory.c which is not built if
> !CONFIG_MMU.
> 
> Regards,
> Axel

Shouldn't might_fault return 0 if !CONFIG_MMU?
No mmu->no fault ...