Hi
It looks like your instruction executed and ready to writeback/commit but
something is blocking it. Check  AtomThread::writeback().
SOM: is the first uop in an x86 instruction
EOM: is the last uop in an x86 instruction
if the x86 instruction has only one uop it will be marked as SOM and EOM
-Furat


On Wed, Oct 30, 2013 at 10:09 AM, Shweta Shinde <[email protected]>wrote:

> Hi,
>
> I am trying to simulate my own instruction in MARSS.
> The instruction does not use any existing uops so I
> also want to add support for my own uops for this instruction.
> I modified qemu decoder to identify this instruction and it is working fine
> without MARSS. When I added support for it in MARSS, things started
> breaking.
> I am able to detect my instruction in decode-complex and add a TransOp
> for it.
> I also added a my new uop to uopimpl.cpp. But my code does not reach this
> uop implementation.
> After debugging I realized that the fetch queue gets filled with the same
> instruction till the assert in AtopmThread::writeback() fails.
>
> Core has not progressed since cycle 5611182 dumping all information
> Atom-Core: 0
>  Fetch Queue:
> Thread: 0 stats:
>  Atom-Ops:
> [ 0] a-op  th 0 uuid          3436596 rip
> 0x000000400ef4 ready-to-writeback
> [nonpipe|] (0 cyc)
> [ 1] a-op  th 0 uuid          3436613 rip
> 0x000000400ef4 ready-to-writeback
> [nonpipe|] (0 cyc)
> ....
> [31]a-op  th 0 uuid          3436625 rip
> 0x000000400ef4 ready-to-writeback
> [nonpipe|] (0 cyc)
>
> I guess I am missing out some details. I have to specify the delay for my
> new uop but I dont know where MARSS does this.
> I also observed in the log that each instruction has a SOM and EOM, but I
> am clueless from where it is coming.
> I would appreciate any help if anyone has tried this before.
>
> Regards,
> Shweta
>
>
> _______________________________________________
> http://www.marss86.org
> Marss86-Devel mailing list
> [email protected]
> https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel
>
_______________________________________________
http://www.marss86.org
Marss86-Devel mailing list
[email protected]
https://www.cs.binghamton.edu/mailman/listinfo/marss86-devel

Reply via email to