[Libre-soc-isa] [Bug 1087] change pseudocode to prevent output register write only when causing a fp trap and output is in same regfile as input
bugzilla-daemon at libre-soc.org
bugzilla-daemon at libre-soc.org
Wed May 24 02:34:20 BST 2023
https://bugs.libre-soc.org/show_bug.cgi?id=1087
--- Comment #21 from Jacob Lifshay <programmerjake at gmail.com> ---
(In reply to Jacob Lifshay from comment #20)
> other *fun* side-effects of fcvtfg
whoops, I meant fcvttgo.
> not writing RT (if we decide to not
> accept my proposal) is that we either have to manually calculate CR0 in the
> pseudocode or not change CR0 (messes with data-dependent fail-first) or
> (probably the worst option) read RT and calculate CR0 from the read-in
> value. (CR1 doesn't have that problem since it's input always comes from
> FPSCR).
>
> I think the best option is to just have fcvtfg
same here
> always write RT as I
> proposed, since suppressing writes (so the trap handler can see the input)
> is only necessary when targeting the same register file as inputs.
--
You are receiving this mail because:
You are on the CC list for the bug.
More information about the Libre-SOC-ISA
mailing list