[Libre-soc-bugs] [Bug 838] sync or at least statically check fields.text, power_decoder, trans/svp64, CSVs between each other

[bugzilla-daemon at libre-soc.org]

https://bugs.libre-soc.org/show_bug.cgi?id=838

--- Comment #15 from Luke Kenneth Casson Leighton <lkcl at lkcl.net> ---
(In reply to Dmitry Selyutin from comment #13)
> 0b0011000000,,,RA,RB,NONE,NONE,NONE,BF,,,,,,,,,,,,,,,cmprb,X,
> 0b1011110011,,,,,RT,,,,,,,,,,,,,,,,,,darn,X,
> 0b1000000000,,,,,,,,,,,,,,,,,,,,,,,mcrxr,X,
> 0b1001000000,,,,,,,,,,,,,,,,,,,,,,,mcrxrx,X,
> 
> These are very, very wrong. Having these in CSV is bad, and, again, the fact
> that these are silently processed and appear to work makes it even worse.
> Any objections if I drop these?

they haven't been properly analysed yet.  so much to do.
sv_analysis.py skips darn.

no absolutely they cannot be removed because the simulator
MUST implement them in scalar.

remember these CSV files are EXECUTABLE not just for SVP64
but Scalar Power ISA v3.0.

so no under NO CIRCUMSTANCES can they be "removed", it will
be as if the instructions had been removed from Scalar Power
ISA 3.0

(In reply to Dmitry Selyutin from comment #14)
> (In reply to Dmitry Selyutin from comment #13)
> > Any objections if I drop these?
> 
> Or, alternatively, please update these with something relevant. I don't
> think leaving the columns empty and handling these cases in code is a good
> idea.

it is what it is.  they cannot be removed, sv_analysis.py has to know
(somehow) and cope with the discernment "what is vectorised and what is not".
likewise mfmsrd, mtmsr, rfid, sc and others cannot and must not be
removed because if they are removed then that catastrophically causes
ISACaller to be unable to execute those *SCALAR* instructions.

-- 
You are receiving this mail because:
You are on the CC list for the bug.