Hi,

Okay, I have been looking at the diff from the patch, but it seems that x86 does not have the file fp64.isa (where the opclass is defined) so I had to use a little the imagination and look for a workaround.


The thing is that the modifications in the file isa_parser.py gave me some idea, so what I did is adding some lines within the condition adding a couple of checks:

elif 'IsFloating' in self.flags:
                if 'div' in self.mnemonic:
                    self.op_class = 'FloatDivOp'
                elif 'mul' in self.mnemonic:
                    self.op_class = 'FloatMultOp'
else:
                    self.op_class = 'FloatAddOp'

As I mentioned, this is just a workaround. I'll try to look for a more elegant fix and update the info as soon as I have it.


Regards,

Ferran O.



On 13/10/16 09:01, Ferran Olid wrote:

Hi,

Thank you very much! I'll try this out right now.


Regards,

Ferran O.

On 13/10/16 02:47, Ayaz Akram wrote:
Hi, you can look at following patch by Fernando, to fix Arm SimdFloat ops labelling:
http://reviews.gem5.org/r/3547/


This will be helpful in doing similar changes for x86.



On Mon, Oct 10, 2016 at 10:19 AM, Ferran Olid <ferran.o...@metempsy.com <mailto:ferran.o...@metempsy.com>> wrote:

    Hi all,

    I am running a code using the x86 decoder and I have just noticed
    that for "addss", "mulss", and "divss" the instructions type
    which is detected by gem5 is FloatAdd (instead of FloatMul and
    FloatDiv for mulss and divss). I attach an example from the debug
    output:

    28189782: system.cpu T0 : @foo+189.0  :   DIVSS_XMM_XMM : mdivf
    %xmm1_low, %xmm1_low, %xmm0_low : FloatAdd : D=0x000000003faf983c


    To me this looks like a bug which is quite important to fix. I
    have been trying to do it though, but with no results yet. Does
    someone have any clue how this can be fixed or if there is any
    existent patch already? I read the documentation in gem5
    (http://www.m5sim.org/X86_Instruction_decoding
    <http://www.m5sim.org/X86_Instruction_decoding>) about the x86
    instruction decode, but it does not quite clarify the code...

    So, any ideas?


    Regards,

    Ferran O.


    _______________________________________________
    gem5-users mailing list
    gem5-users@gem5.org <mailto:gem5-users@gem5.org>
    http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
    <http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users>




_______________________________________________
gem5-users mailing list
gem5-users@gem5.org
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users



_______________________________________________
gem5-users mailing list
gem5-users@gem5.org
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users

_______________________________________________
gem5-users mailing list
gem5-users@gem5.org
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users

Reply via email to