Re: [PATCH] MIPS: Loongson64: Give chance to build under !CONFIG_NUMA and !CONFIG_SMP

[Tiezhu Yang]

On 12/15/2020 09:21 PM, Thomas Bogendoerfer wrote:
> On Thu, Dec 03, 2020 at 08:32:52PM +0800, Tiezhu Yang wrote:
> > In the current code, we can not build under !CONFIG_NUMA and !CONFIG_SMP
> > on the Loongson64 platform, it seems bad for the users who just want to
> > use pure single core (not nosmp) to debug, so do the following things to
> > give them a chance:
> >
> > (1) Do not select NUMA and SMP for MACH_LOONGSON64 in Kconfig, make NUMA
> > depends on SMP, and then just set them in the loongson3_defconfig.
> > (2) Move szmem() from numa.c to init.c and add prom_init_memory() under
> > !CONFIG_NUMA.
> > (3) Clean up szmem() due to the statements of case SYSTEM_RAM_LOW and
> > SYSTEM_RAM_HIGH are the same.
> > (4) Remove the useless declaration of prom_init_memory() and add the
> > declaration of szmem() in loongson.h to avoid build error.
> >
> > Signed-off-by: Youling Tang <tangyouling@xxxxxxxxxxx>
> > Signed-off-by: Jinyang He <hejinyang@xxxxxxxxxxx>
> > Signed-off-by: Tiezhu Yang <yangtiezhu@xxxxxxxxxxx>
> > ---
> >   arch/mips/Kconfig                                |  3 +-
> >   arch/mips/configs/loongson3_defconfig            |  2 +
> >   arch/mips/include/asm/mach-loongson64/loongson.h |  2 +-
> >   arch/mips/loongson64/init.c                      | 49 ++++++++++++++++++++++
> >   arch/mips/loongson64/numa.c                      | 52 +-----------------------
> >   5 files changed, 54 insertions(+), 54 deletions(-)
> >
> > diff --git a/arch/mips/Kconfig b/arch/mips/Kconfig
> > index 44a47ad..2034c66 100644
> > --- a/arch/mips/Kconfig
> > +++ b/arch/mips/Kconfig
> > @@ -490,8 +490,6 @@ config MACH_LOONGSON64
> >   	select SYS_SUPPORTS_ZBOOT
> >   	select SYS_SUPPORTS_RELOCATABLE
> >   	select ZONE_DMA32
> > -	select NUMA
> > -	select SMP
> >   	select COMMON_CLK
> >   	select USE_OF
> >   	select BUILTIN_DTB
> > @@ -2755,6 +2753,7 @@ config ARCH_SPARSEMEM_ENABLE
> >   config NUMA
> >   	bool "NUMA Support"
> >   	depends on SYS_SUPPORTS_NUMA
> > +	depends on SMP
> can you solve your problem without this hunk ? I don't want to make NUMA
> depeding on SMP. NUMA just selects memory archtitecture.

Hi Thomas,

I have tested the following three configs on the Loongson platform:
(1) !NUMA and !SMP,
(2) !NUMA and SMP,
(3) NUMA and SMP,
everything is all right.

But there exists the following build error under NUMA and !SMP:

  CC      arch/mips/kernel/asm-offsets.s
In file included from ./include/linux/gfp.h:9:0,
                 from ./include/linux/xarray.h:14,
                 from ./include/linux/radix-tree.h:19,
                 from ./include/linux/fs.h:15,
                 from ./include/linux/compat.h:17,
                 from arch/mips/kernel/asm-offsets.c:12:
./include/linux/topology.h: In function ‘numa_node_id’:
./include/linux/topology.h:119:2: error: implicit declaration of 
function ‘cpu_logical_map’ [-Werror=implicit-function-declaration]
  return cpu_to_node(raw_smp_processor_id());
  ^
cc1: some warnings being treated as errors
scripts/Makefile.build:117: recipe for target 
'arch/mips/kernel/asm-offsets.s' failed
make[1]: *** [arch/mips/kernel/asm-offsets.s] Error 1
arch/mips/Makefile:396: recipe for target 'archprepare' failed
make: *** [archprepare] Error 2

I find a patch to fix this kind of build errors [1], but it seems
meaningless.

According to the NUMA and SMP description in arch/mips/Kconfig,
we will use only one CPU of a multiprocessor machine if !SMP,
on single node systems leave NUMA disabled.

So I think there is no need to use NUMA if !SMP, and also we should
make NUMA depend on SMP to avoid build errors.

Thanks,
Tiezhu

[1] MIPS: Loongson: Fix complie errors without CONFIG_SMP
https://lore.kernel.org/patchwork/patch/1295704/

> Thomas.