Add F/D to Zcf/d's depending relations #224
Conversation
|
This change is positive for me from an LLVM perspective - having zcf and zcd imply F/D seems more consistent with other extensions, and we'd like to update the behaviour of LLVM if this change is committed. |
|
I think it makes sense, added @abukharmeh and @aswaterman to get their opinions |
|
It makes sense to me, but it is a change. I think you could treat it as a clarification, as readers might've reasonably inferred these dependences, but mention it to tech-unprivileged to make sure no one has actually built an implementation that took the old spec literally. |
|
emailed to the mailing list: https://lists.riscv.org/g/tech-unprivileged/message/519 |
Co-authored-by: Nicolas Brunie <[email protected]> Signed-off-by: Tariq Kurd <[email protected]>
|
No feedback from the mailing list so far - are we happy for this to go through as a clarification, or wait a bit longer? |
|
IMO, give it until the end of the week, then merge. |
This was an omission in the spec that has now been addressed riscvarchive/riscv-code-size-reduction#224. Differential Revision: https://reviews.llvm.org/D156314
This was an omission in the spec that has now been addressed riscvarchive/riscv-code-size-reduction#224. Differential Revision: https://reviews.llvm.org/D156314
This was an omission in the spec that has now been addressed riscvarchive/riscv-code-size-reduction#224. Differential Revision: https://reviews.llvm.org/D156314
Add F/D to Zcf/d's depending relations, since Zcf/d need F/D extension to generate instructions as Zca sections said.
#221 (comment)