alexcrichton commented on Issue #1620:
r=me on the wasmtime bits here, thanks!
github-actions[bot] commented on Issue #1620:
Subscribe to Label Action
cc @bnjbvr, @peterhuene
<details>
This issue or pull request has been labeled: "cranelift", "wasmtime:api"Thus the following users have been cc'd because of the following labels:
- bnjbvr: cranelift
- peterhuene: wasmtime:api
To subscribe or unsubscribe from this label, edit the <code>.github/subscribe-to-label.json</code> configuration file.
Learn more.
</details>
stuaxo commented on Issue #1620:
This is awesome, could it help Rust compile dos executables ?
whitequark commented on Issue #1620:
This is awesome, could it help Rust compile dos executables ?
Certainly not, this PR is not even remotely related to DOS and I cannot imagine that Cranelift would ever be able to emit real-mode x86 machine code.
bjorn3 commented on Issue #1620:
Cranelift could theoretically add
data32
prefixes everywhere.
stuaxo commented on Issue #1620:
Oh heh, I meant 32 bit protected mode DOS, but fair enough.
whitequark commented on Issue #1620:
I'd give ArgumentPurpose::CalleeSaved a shot here, this is well after register allocation so it's probably free of risk in side effects there.
I tried that first and I think it either didn't work correctly or generated useless instructions. I can recheck it if you want to be sure, but that was my first implementation.
iximeow commented on Issue #1620:
I'd greatly appreciate if you could check again. If it fails in a weird way I'd think Cranelift should be better-behaved. Worst case, it helps motivate an atypical ArgumentPurpose that isn't
CalleeSave
. Or, we should be clearer about the semantics ofArgumentPurpose
andNormal
should be acceptable (by being an argument to the function, even if it's not an ABI-mandated one)
whitequark commented on Issue #1620:
@iximeow I stand corrected--there seems to be no observable difference and everything apperas to work just fine with %rsp as CSR.
whitequark edited a comment on Issue #1620:
@iximeow I stand corrected--there seems to be no observable difference in the generated code and everything appears to work just fine with %rsp as CSR.
whitequark commented on Issue #1620:
I've updated the PR to use
ArgumentPurpoe::CalleeSave
.
Last updated: Dec 23 2024 at 12:05 UTC