Re: [Capstone-users] Adding regs_access support for PPC and MIPS

[Derrick M. <der...@gm...>]

The idea is to make the generation of the needed inc file less of a black
box, and ensure that it is properly ordered.

On Wed, Jul 31, 2019, 06:42 Nguyen Anh Quynh <aq...@gm...> wrote:

>
>
> On Tue, Jul 30, 2019, 23:07 Derrick McKee <der...@gm...> wrote:
>
>> So, looking closer at the existing MappingInsnOp.inc files, it looks like
>> my proposed YAML format is not suitable. So instead I propose the following:
>>
>> output: "MipMappingInsnOp.inc"
>> Insns:
>> - enum: MIPS_INS_ABSQ_S
>>   subtypes:
>>     - examples: {
>>       Mips_ABSQ_S_PH: "ABSQ_S.PH  rd, rt",
>>       Mips_ABSQ_S_QB: "ABSQ_S.QB rd, rt",
>>       Mips_ABSQ_S_W: "ABSQ_S.W  rd, rt"
>>     }
>>       operands:
>>         - read: No
>>           write: Yes
>>         - read: Yes
>>           write: No
>>
>
> Yes sounds good.
>
> But how does this make it better than current .inc format? A bit more
> readability, but not too much to me.
>
>
>
>>
>> On Tue, Jul 30, 2019 at 8:45 AM Derrick McKee <der...@gm...>
>> wrote:
>>
>>> I think I am mostly done with the C implementation, and have updated the
>>> bindings. I just need to generate the {Mips,PPC}MappingInsnOp.inc files.
>>> The existing files for ARM and x86 seem to contain the read/write
>>> information for every instruction, sorted by enum name.  They also say that
>>> they were autogenerated, but I guess that isn't quite correct.  I am
>>> thinking that I can write a Python script that reads in a YAML file, and
>>> outputs the properly sorted inc file containing all the information
>>> (including comments) currently in existing InsnOp.inc files.  The YAML file
>>> will have a structure like the following:
>>>
>>> output: "MipsMappingInsnOp.inc"
>>> Insns:
>>>     - enum: MIPS_INS_ADD
>>>       example: "add $d, $t, $s"
>>>       operands:
>>>           - read: No
>>>             write: Yes
>>>           - read: Yes
>>>             write: No
>>>           - read: Yes
>>>             write: No
>>>
>>> Thoughts?
>>>
>>> On Tue, Jul 30, 2019 at 1:24 AM Nguyen Anh Quynh <aq...@gm...>
>>> wrote:
>>>
>>>> Those .inc files involved a lot of manual work, that is why it takes
>>>> time.
>>>>
>>>> What is your progress?
>>>>
>>>> Cheers.
>>>>
>>>> On Mon, Jul 29, 2019, 23:47 Derrick McKee <der...@gm...>
>>>> wrote:
>>>>
>>>>> Hi,
>>>>>
>>>>> I am beginning to add support for regs_access for MIPS and PPC.
>>>>> Currently, I am following what has been done for ARM and x86, which as far
>>>>> as I can tell for PPC involves adding a new access member to the
>>>>> cs_ppc_op struct, writing a new PPC_regs_access function, and setting
>>>>> the regs_access function pointer in cs_struct in PPC_global_init.  My
>>>>> question is how is the access member set?  For both ARM and x86, there is a
>>>>> MappingInsnOp.inc file that appears to be autogenerated, which contains the
>>>>> access information.  How was this file generated?
>>>>>
>>>>> --
>>>>> Derrick McKee
>>>>> Phone: (703) 957-9362
>>>>> Email: der...@gm...
>>>>> _______________________________________________
>>>>> Capstone-users mailing list
>>>>> Cap...@li...
>>>>> https://lists.sourceforge.net/lists/listinfo/capstone-users
>>>>>
>>>> _______________________________________________
>>>> Capstone-users mailing list
>>>> Cap...@li...
>>>> https://lists.sourceforge.net/lists/listinfo/capstone-users
>>>>
>>>
>>>
>>> --
>>> Derrick McKee
>>> Phone: (703) 957-9362
>>> Email: der...@gm...
>>>
>>
>>
>> --
>> Derrick McKee
>> Phone: (703) 957-9362
>> Email: der...@gm...
>> _______________________________________________
>> Capstone-users mailing list
>> Cap...@li...
>> https://lists.sourceforge.net/lists/listinfo/capstone-users
>>
> _______________________________________________
> Capstone-users mailing list
> Cap...@li...
> https://lists.sourceforge.net/lists/listinfo/capstone-users
>