Wasmtime GitHub notifications bot (Sep 12 2024 at 20:56): Wasmtime GitHub notifications bot (Sep 12 2024 at 20:56):fitzgen requested wasmtime-default-reviewers for a review on PR #9235.
Wasmtime GitHub notifications bot (Sep 12 2024 at 20:56):fitzgen requested pchickey for a review on PR #9235.
Wasmtime GitHub notifications bot (Sep 12 2024 at 20:56):fitzgen opened PR #9235 from fitzgen:pulley-compare-and-branch-64 to bytecodealliance:main:
Need these for a follow up PR.
<!--
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
-->
Wasmtime GitHub notifications bot (Sep 12 2024 at 20:56):fitzgen requested alexcrichton for a review on PR #9235.
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:00):fitzgen updated PR #9235.
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:01):alexcrichton created PR review comment:
Not necessary for this PR, but I think it'd be good to avoid hardcoding 7 here eventually. This'll be hard to remember to update if we pack the
a/bregs into a single byte for example (not that that's possible because there's 32-registers but similar-ish other encoding schemes might happen)
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:01):alexcrichton submitted PR review.
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:01):alexcrichton created PR review comment:
To bikeshed the names about, how about using the CLIF names which are all generally three letters? For example
neandeqfor the first two andslt,sle,ult, andulefor the latter four.
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:03):fitzgen submitted PR review.
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:03):fitzgen created PR review comment:
These names match the 32-bit variants and the comparison-without-branching instructions. We can rename everything in a follow up, if you want, but I won't do it in this PR.
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:05):fitzgen submitted PR review.
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:05):fitzgen created PR review comment:
We can certainly define a
const OFFSET_OF_JUMP_IN_COND_BR_INST: usize = 7constant and use that in place of this, but I don't know how we could get the macro to automatically keep the value up-to-date, so I'm not sure it would actually change things in practice. We'd still need to remember to update the constant definition when we bitpack stuff, for example.
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:05):alexcrichton submitted PR review.
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:05):alexcrichton created PR review comment:
Sounds good :+1:
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:06):alexcrichton submitted PR review.
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:06):alexcrichton created PR review comment:
I forget if this was already tried, but could relocation offsets be done such that a static offset here isn't needed? I forget if it's relative to the start or the end of the instruction but if we flip it then this constant here would go away and it's just the producer, Cranelift, that would get a bit more complicated right?
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:08):fitzgen updated PR #9235.
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:09):fitzgen has enabled auto merge for PR #9235.
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:11):fitzgen submitted PR review.
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:11):fitzgen created PR review comment:
Yeah, right now the jump is relative from the relative-offset's position, not the instruction's position. We could indeed make it relative to the instruction's position and then tweak emission in Cranelift a bit. That would be kind of nice, one less ALU op at runtime!
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:31):fitzgen requested wasmtime-compiler-reviewers for a review on PR #9235.
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:31):fitzgen updated PR #9235.
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:31):fitzgen requested cfallin for a review on PR #9235.
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:40):fitzgen has enabled auto merge for PR #9235.
Wasmtime GitHub notifications bot (Sep 12 2024 at 21:59):fitzgen merged PR #9235.
Last updated: Oct 23 2024 at 20:03 UTC