The instResult is used in the InOrder model but not in the O3 as I could tell.
I plan to merge the BaseDynInst object for the O3 and InOrder and the InstResult will probably get whacked. On Tue, May 10, 2011 at 4:03 PM, Gabriel Michael Black < [email protected]> wrote: > As far as I know instResult is a historic relic. You should be able to just > look at the physical registers. > > Gabe > > > Quoting Andrea Pellegrini <[email protected]>: > > >> Hi all, >> I am working with X86 on M5, and I am trying to track the values that are >> written back to the RF when an instructions commits. Some X86 instructions >> have multiple destination registers, for instance: >> >> 182500: system.cpu.commit: apellegr TGcommit CSV, 1 , >> (0x400193=>0x40019b).(0=> >> 4194715), XOR_R_R : xor rbp, rbp, rbp, >> apellegr Will write on register: 5 >> apellegr Will write on register: 32 >> >> However, in the BaseDynInst there is only a single field for the integer >> result: >> >> Result instResult; >> >> and in the execution logic, the index of the destination register is not >> even used: >> >> -------- >> >> void BaseO3DynInst::setIntRegOperand(const StaticInst *si, int idx, >> uint64_t val) >> { >> this->cpu->setIntReg(this->_destRegIdx[idx], val); >> BaseDynInst<Impl>::setIntRegOperand(si, idx, val); >> } >> >> -------- >> >> void BaseDynInst<Impl>::setIntRegOperand(const StaticInst *si, int idx, >> uint64_t val) >> { >> if (recordResult) >> instResult.integer = val; >> } >> >> ---------------------------- >> >> My question is: >> Can I just look at the value of the physical register assigned to retrieve >> the correct value of the destination register? >> Is the "instResult" field redundant? >> >> Thanks >> -Andrea >> _______________________________________________ >> m5-users mailing list >> [email protected] >> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users >> >> > > _______________________________________________ > m5-users mailing list > [email protected] > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users > -- - Korey
_______________________________________________ m5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
