Illegal Instruction Exception not Raised - FS Field

[shetalani]

RISC-V Specification:

• "The FS field encodes the status of the floating-point unit, including the CSR fcsr and floating-point data registers f0–f31"
• "In systems that do not implement S-mode and do not have a floating-point unit, the FS field is hardwired to zero."
• "When an extension’s status is set to Off, any instruction that attempts to read or write the corresponding state will cause an illegal instruction exception."

Issue Description:

Accessing the F-extension CSRs / floating-point data registers f0–f31 while the FS field of MSTATUS is set to OFF doesn't raise an illegal instruction exception.

Example:

As shown below, the instruction 32'h23aaf3 (csrrs x21, frm, x7) is decoded at time point t##0, while FS field is set to OFF, with no illegal instruction being flagged, as illegal_insn_dec is de-asserted.

---

Product: OneSpin 360 DV-Verify
App: RVV
Tool's version: 2019.2.2

[EkanshBhatnagar]

Can I tackle this bug ?

[davideschiavone]

hi @EkanshBhatnagar , of course you can, everyone can :) thanks, please synchronize with @pascalgouedo

[pascalgouedo]

Hi,
After some CV32E40P RTL analysis, here is the status:

• FS (and SD) are always hardwired to 0 even when FPU = 1

After some RISC-V spec analysis, here are some remarks:

• If no S-Mode and FPU = 0 (v1 tag case), F-extension CSRs / f0–f31 registers read/write accesses and FS field write access should raise illegal exception => agreed.
• If no S-Mode and FPU = 1 (will be v2 tag case), FS should be writeable by csr instruction and modifiable to Dirty when F-extension CSRs / f0–f31 registers are written/updated => Not done in present v1 RTL. New issue to create.

[pascalgouedo]

Resolved with PR #801