Re: [PATCH 26/31] nds32: Build infrastructure

[Greentime Hu]

2017-11-13 18:45 GMT+08:00 Geert Uytterhoeven <geert@xxxxxxxxxxxxxx>:
> On Wed, Nov 8, 2017 at 11:16 AM, Arnd Bergmann <arnd@xxxxxxxx> wrote:
>> On Wed, Nov 8, 2017 at 6:55 AM, Greentime Hu <green.hu@xxxxxxxxx> wrote:
>>> --- /dev/null
>>> +++ b/arch/nds32/Kconfig
>>> @@ -0,0 +1,107 @@
>>> +#
>>> +# For a description of the syntax of this configuration file,
>>> +# see Documentation/kbuild/kconfig-language.txt.
>>> +#
>>> +
>>> +config NDS32
>>> +        def_bool y
>>> +       select ARCH_HAS_RAW_COPY_USER
>>> +       select ARCH_WANT_FRAME_POINTERS if FTRACE
>>> +       select ARCH_WANT_IPC_PARSE_VERSION
>>> +       select CLKSRC_MMIO
>>> +       select CLONE_BACKWARDS
>>> +       select TIMER_OF
>>> +       select FRAME_POINTER
>>> +       select GENERIC_ATOMIC64
>>> +       select GENERIC_CPU_DEVICES
>>> +       select GENERIC_CLOCKEVENTS
>>> +       select GENERIC_IOMAP
>>> +       select GENERIC_IRQ_CHIP
>>> +       select GENERIC_IRQ_PROBE
>>> +       select GENERIC_IRQ_SHOW
>>> +       select GENERIC_STRNCPY_FROM_USER
>>> +       select GENERIC_STRNLEN_USER
>>> +       select GENERIC_TIME_VSYSCALL
>>> +       select HAVE_ARCH_TRACEHOOK
>>> +       select HAVE_GENERIC_IOMAP
>>
>> You normally don't want HAVE_GENERIC_IOMAP, at least unless the CPU
>> has special instructions to trigger PCI I/O port access.
>
> GENERIC_IOMAP or HAVE_GENERIC_IOMAP?
>
> The latter doesn't even exist, except in
> Documentation/kbuild/kconfig-language.txt, so it cannot be selected.

Thanks.
I will remove HAVE_GENERIC_IOMAP in the next version patch.

> Given the checks for __NDS32_EB__, NDS32 can be either big or little endian,
> so you should have (excatly one of) CPU_BIG_ENDIAN or CPU_LITTLE_ENDIAN set.

Thanks.
I will check if we need this config or not and update in the next version patch.