[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [RFC v3 1/4] softmmu: move softmmu only files from root
From: |
Philippe Mathieu-Daudé |
Subject: |
Re: [RFC v3 1/4] softmmu: move softmmu only files from root |
Date: |
Mon, 25 May 2020 19:32:49 +0200 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Thunderbird/68.5.0 |
On 5/25/20 5:37 PM, Claudio Fontana wrote:
> On 5/25/20 5:12 PM, Philippe Mathieu-Daudé wrote:
>> On 5/25/20 4:54 PM, Claudio Fontana wrote:
>>> move arch_init, balloon, cpus, ioport, memory, memory_mapping, qtest.
>>>
>>> They are all specific to CONFIG_SOFTMMU.
>>>
>>> Signed-off-by: Claudio Fontana <address@hidden>
>>> ---
>>> MAINTAINERS | 12 ++++++------
>>> Makefile.target | 7 ++-----
>>> softmmu/Makefile.objs | 10 ++++++++++
>>> arch_init.c => softmmu/arch_init.c | 0
>>
>> OK.
>>
>>> balloon.c => softmmu/balloon.c | 0
>>
>> OK
>>
>>> cpus.c => softmmu/cpus.c | 0
>>
>> I'm still not 100% convinced.
>
> Hi Philippe, do you see a reason to keep cpus.c in the top_srcdir?
>
> The way I see it, it belongs in softmmu/,
>
> we have a separate cpu-common.c that is target-indep and in common between
> user and softmmu,
> while cpus.c is softmmu-only right?
Sorry, I meant to say I need to look more in details at this file uses
and will do a second review pass on your series.
>
>
>>
>>> ioport.c => softmmu/ioport.c | 0
>>
>> OK
>>
>>> memory.c => softmmu/memory.c | 0
>>
>> OK
>>
>>> memory_mapping.c => softmmu/memory_mapping.c | 0
>>
>> I am not sure (yet).
>
> headers are already in sysemu (sysemu/memory_mapping.h)
>
>>
>>> qtest.c => softmmu/qtest.c | 0
>>
>> This one seems to belong to accel/
>
> We already have accel/qtest.c , that is the qtest AccelClass.
>
> Here we are talking about qtest.c from the top_srcdir, which it the
> qtest server and is built with softmmu code, with header in sysemu/qtest.h
I haven't looked at it in detail but will.
>
> Thomas?
>
>>> 10 files changed, 18 insertions(+), 11 deletions(-)
>>> rename arch_init.c => softmmu/arch_init.c (100%)
>>> rename balloon.c => softmmu/balloon.c (100%)
>>> rename cpus.c => softmmu/cpus.c (100%)
>>> rename ioport.c => softmmu/ioport.c (100%)
>>> rename memory.c => softmmu/memory.c (100%)
>>> rename memory_mapping.c => softmmu/memory_mapping.c (100%)
>>> rename qtest.c => softmmu/qtest.c (100%)
>>
>> What about the corresponding headers?
>
> headers already in sysemu/ :
>
> sysemu/arch_init.h
> sysemu/balloon.h
> sysemu/cpus.h
> sysemu/qtest.h - (again, for the qtest server)
> sysemu/memory_mapping.h
>
> headers currently in exec/ :
>
> exec/memory.h
> exec/ioport.h
Yes, if you want a clean separation, I'd rather restrict
include/softmmu/ to softmmu builds. user-mode and tools builds shouldn't
access these headers.
>
> Paolo?
>
>>
>> [...]
>>
>
- [RFC v3 0/4] QEMU cpus.c refactoring, Claudio Fontana, 2020/05/25
- [RFC v3 2/4] cpu-throttle: new module, extracted from cpus.c, Claudio Fontana, 2020/05/25
- [RFC v3 3/4] cpu-timers, icount: new modules, Claudio Fontana, 2020/05/25
- [RFC v3 4/4] cpus: extract out accel-specific code to each accel, Claudio Fontana, 2020/05/25