Rather than having a different opcode for RV32 and RV64. Let's just say the integer type is XLenVT and use a single opcode for both modes.
Details
Diff Detail
Event Timeline
Makes sense to me otherwise but I'll defer my LGTM to someone else.
llvm/lib/Target/RISCV/RISCVISelLowering.h | ||
---|---|---|
56–57 | nit: the comments for these notes are still third-person plural despite each being a single node: match and are; I think matches and is is better. |
llvm/lib/Target/RISCV/RISCVInstrInfoZfh.td | ||
---|---|---|
344 | Hm, looking at the other files this really should have had a blank line above it, which is more important now you're adding a new set of patterns to the conversion operations "section". | |
348 | This could do with a comment; F has // Moves (no conversion) for its bitcasts. |
Ah, I hadn't realised the F comment got lost in 741b04b0b7912611a8a5b7e74462e87b8930a116, I was looking at our fork which is currently still based on 11. This looks good to me now.
I'm not sure why I dropped it unless I was basing it on the lack of comment for the RV64 patterns.
nit: the comments for these notes are still third-person plural despite each being a single node: match and are; I think matches and is is better.