vulc41n requested abrown for a review on PR #9114.
vulc41n requested wasmtime-compiler-reviewers for a review on PR #9114.
vulc41n opened PR #9114 from vulc41n:winch-aarch64-extend-wrap
to bytecodealliance:main
:
Hey :wave:
This PR implements
extend
,wrap
,popcnt
,promote
&demote
instructions for winch targeting aarch64.#8321
<!--
Please make sure you include the following information:
If this work has been discussed elsewhere, please include a link to that
conversation. If it was discussed in an issue, just mention "issue #...".Explain why this change is needed. If the details are in an issue already,
this can be brief.Our development process is documented in the Wasmtime book:
https://docs.wasmtime.dev/contributing-development-process.htmlPlease ensure all communication follows the code of conduct:
https://github.com/bytecodealliance/wasmtime/blob/main/CODE_OF_CONDUCT.md
-->
vulc41n requested alexcrichton for a review on PR #9114.
vulc41n requested wasmtime-core-reviewers for a review on PR #9114.
vulc41n updated PR #9114.
vulc41n updated PR #9114.
saulecabrera commented on PR #9114:
I can help reviewing this one.
saulecabrera requested saulecabrera for a review on PR #9114.
saulecabrera submitted PR review.
saulecabrera created PR review comment:
To reduce register pressure, could we use the scratch float register here?
saulecabrera created PR review comment:
In the case of
aarch64
I believe that thefallback
test is redundant i.e., since we are not generating an alternative sequence of instructions like in x64.
saulecabrera created PR review comment:
If I'm not wrong, Wasm only defined
{i32, i64}_popcnt
so I believe we only need to handle theS32 | S64
case and handle all the other operand sizes withunreachable!()
instead ofunimplemented
, which also means that we can probably drop theaddp_rrr
implementation.
saulecabrera created PR review comment:
Could we add some methods to the
ExtendKind
implementation instead? Something like:impl ExtendKind { fn signed(&self) -> bool { ... } fn from_bits(&self) -> u8 { ... } fn to_bits(&self) -> u8 { ... } }
saulecabrera edited PR review comment.
vulc41n submitted PR review.
vulc41n created PR review comment:
You're right, I used
lower.isle
as a reference but CLIF is different from WASM.handle all the other operand sizes with
unreachable!()
instead ofunimplemented
I just removed the match, please tell me if you think an assertion is needed
vulc41n updated PR #9114.
saulecabrera submitted PR review:
One minor nit and this should be good to go.
saulecabrera submitted PR review:
One minor nit and this should be good to go.
saulecabrera created PR review comment:
The scratch register is manually tracked i.e.,
free_reg
is a no-op in this case, so we can remove this call.
vulc41n updated PR #9114.
vulc41n updated PR #9114.
saulecabrera submitted PR review:
Thanks!
saulecabrera merged PR #9114.
Last updated: Nov 22 2024 at 16:03 UTC