[libre-riscv-dev] sv.setvl encoding

Luke Kenneth Casson Leighton lkcl at lkcl.net
Fri Jun 28 11:08:51 BST 2019

On Fri, Jun 28, 2019 at 10:05 AM Luke Kenneth Casson Leighton
<lkcl at lkcl.net> wrote:

>  ah maaaan, i have to back out most of the changes i added to spike-sv
> :)  the dest-subvl-offs is just an independent variable that needs to
> be FSM-looped-on, it's *not* treated the same way as src-pred
> dest-pred at all.

ok done (whew), bit of a scary moment where i had been trying to put
in mret / sret svstatus restoration at the same time, and broke stuff
(whoops).  set on the SVSTATE CSR is definitely broken.

once that was out the way (gave up, commented-out the CSRRW on
SVSTATE), i managed to morph the code, introducing just a single
FSM-based loop on setvl with a single argument rather than 2.

subvl unit tests (including predicated addi) still work, and so do the
other unit tests (whew)


More information about the libre-riscv-dev mailing list