Re: [Qemu-devel] [PATCH] tci: Fix compile failure by including qemu-common.h

[Peter Crosthwaite <peter.crosthwaite@xilinx.com>]

On Thu, Jul 9, 2015 at 9:58 AM, Peter Maydell <peter.maydell@linaro.org> wrote:
> On 9 July 2015 at 17:49, Peter Maydell <peter.maydell@linaro.org> wrote:
>> Compilation of TCI was accidentally broken by the recent disassembler
>> changes:
>>
>>   CC    x86_64-softmmu/arch_init.o
>> In file included from target-i386/cpu-qom.h:23:0,
>>                  from target-i386/cpu.h:986,
>>                  from include/qemu-common.h:122,
>>                  from include/disas/bfd.h:12,
>>                  from disas/tci.c:20:
>> include/qom/cpu.h:178:43: error: unknown type name ‘disassemble_info’
>>      void (*disas_set_info)(CPUState *cpu, disassemble_info *info);
>>                                            ^
>> include/qom/cpu.h:179:1: error:
>> no semicolon at end of struct or union [-Werror]
>>  } CPUClass;
>>  ^
>> cc1: all warnings being treated as errors
>>
>> The underlying cause of this is an include loop:
>>  bfd.h -> qemu-common.h -> target-arm/cpu.h ->  target-arm/cpu-qom.h
>>   -> qom/cpu.h -> bfd.h
>>
>> which means that if bfd.h is included first then qom/cpu.h doesn't
>> get the definition of the disassemble_info type that it wanted.
>> The easiest fix for this is to include qemu-common.h from tci.c
>> before including disas/bfd.h.
>>
>> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
>
> Applied to master.
>

Thanks,

I'll be adding this one to my own configury for multi-arch testing.

Regards,
Peter

> Note that "make check" on a TCI seems to assert; I'll let
> somebody who cares about TCI look into that.
>
> -- PMM
>