[PATCH 1/2] powerpc: drop dependency on <asm/machdep.h> in archrandom.h

[Michael Ellerman]

Yury Norov <yury.norov at gmail.com> writes:
> On Mon, Jul 25, 2022 at 10:22:13PM +1000, Michael Ellerman wrote:
>> Michael Ellerman <mpe at ellerman.id.au> writes:
>> > Yury Norov <yury.norov at gmail.com> writes:
>> >> archrandom.h includes <asm/machdep.h> to refer ppc_md. This causes
>> >> circular header dependency, if generic nodemask.h  includes random.h:
>> >>
>> >> In file included from include/linux/cred.h:16,
>> >>                  from include/linux/seq_file.h:13,
>> >>                  from arch/powerpc/include/asm/machdep.h:6,
>> >>                  from arch/powerpc/include/asm/archrandom.h:5,
>> >>                  from include/linux/random.h:109,
>> >>                  from include/linux/nodemask.h:97,
>> >>                  from include/linux/list_lru.h:12,
>> >>                  from include/linux/fs.h:13,
>> >>                  from include/linux/compat.h:17,
>> >>                  from arch/powerpc/kernel/asm-offsets.c:12:
>> >> include/linux/sched.h:1203:9: error: unknown type name 'nodemask_t'
>> >>  1203 |         nodemask_t                      mems_allowed;
>> >>       |         ^~~~~~~~~~
>> >>
>> >> Fix it by removing <asm/machdep.h> dependency from archrandom.h
>> >>
>> >> Signed-off-by: Yury Norov <yury.norov at gmail.com>
>> >> ---
>> >>  arch/powerpc/include/asm/archrandom.h |  9 +--------
>> >>  arch/powerpc/kernel/setup-common.c    | 11 +++++++++++
>> >>  2 files changed, 12 insertions(+), 8 deletions(-)
>> >>
>> >> diff --git a/arch/powerpc/include/asm/archrandom.h b/arch/powerpc/include/asm/archrandom.h
>> >> index 9a53e29680f4..21def59ef1a6 100644
>> >> --- a/arch/powerpc/include/asm/archrandom.h
>> >> +++ b/arch/powerpc/include/asm/archrandom.h
>> >> @@ -4,7 +4,7 @@
>> >>  
>> >>  #ifdef CONFIG_ARCH_RANDOM
>> >>  
>> >> -#include <asm/machdep.h>
>> >> +bool __must_check arch_get_random_seed_long(unsigned long *v);
>> >>  
>> >>  static inline bool __must_check arch_get_random_long(unsigned long *v)
>> >>  {
>> >> @@ -16,13 +16,6 @@ static inline bool __must_check arch_get_random_int(unsigned int *v)
>> >>  	return false;
>> >>  }
>> >>  
>> >> -static inline bool __must_check arch_get_random_seed_long(unsigned long *v)
>> >> -{
>> >> -	if (ppc_md.get_random_seed)
>> >> -		return ppc_md.get_random_seed(v);
>> >> -
>> >> -	return false;
>> >> -}
>> >
>> > I'd rather we didn't have to force this out of line.
>> >
>> > I think I see a different way to fix it, I'll just do some more build
>> > tests.
>> 
>> Of course my idea didn't work :}
>> 
>> So I'll just ack your patch for now, and maybe I can get the headers
>> cleaned up in future to allow it to be out-of-line again.
>
> I understand that it looks like a tradeoff - we inline a couple of small
> functions with the cost of uninlining an almost innocent victim. 

Yeah. The truth is the cost to access the RNG will far outweigh the cost
of that out-of-line call, so there's no real issue. But it's also such a
small function that it just cries out to be inlined :)

> The complete solution would be probably a splitting ppc_md declaration
> out of asm/machdep.h. I wanted to do that, but I'm not a PPC guy, and
> just don't know how to split the header correctly.

I managed to drop the includes of seq_file.h and dma-mapping.h, which
seemed to fix the circular include problem, but there's a bit of fall
out in unrelated files. I think I can get that sorted though eventually.

> Anyways, thanks for the ack. Applied on bitmap-for-next.

No worries.

cheers