cfallin commented on issue #6384:
(Assigning myself -- talked with Alex earlier today and happy to take this one)
github-actions[bot] commented on issue #6384:
Subscribe to Label Action
cc @fitzgen, @peterhuene, @saulecabrera
<details>
This issue or pull request has been labeled: "cranelift", "cranelift:area:aarch64", "cranelift:area:machinst", "cranelift:area:x64", "cranelift:meta", "cranelift:module", "fuzzing", "wasmtime:api", "winch"Thus the following users have been cc'd because of the following labels:
- fitzgen: fuzzing
- peterhuene: wasmtime:api
- saulecabrera: winch
To subscribe or unsubscribe from this label, edit the <code>.github/subscribe-to-label.json</code> configuration file.
Learn more.
</details>
afonso360 commented on issue #6384:
Hey, I was playing around with adding this setting to fuzzgen and it found a panic on AArch64, looks like when setting
bb_padding_log2
to 1, (I suspect) we get an unaligned basic block, which causes issues.<details>
<summary>Testcase</summary>
test compile set bb_padding_log2=1 target aarch64 function u1:0(i64) -> i64{ block0(v0: i64): jump block1 block1: return v0 }
Panic:
thread 'worker #0' panicked at 'assertion failed: pc_rel & 3 == 0', cranelift/codegen/src/isa/aarch64/inst/mod.rs:2847:17 note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace ERROR cranelift_filetests::concurrent > FAIL: panicked in worker #0: assertion failed: pc_rel & 3 == 0 FAIL ./lmao.clif: panicked in worker #0: assertion failed: pc_rel & 3 == 0 1 tests Error: 1 failure
</details>
alexcrichton commented on issue #6384:
we get an unaligned basic block, which causes issues.
Indeed! Fuzzing for me locally turned that up quite quickly and I forgot to push that up last night. I added a re-alignment now to get past that. Fuzzing found another panic last night so I'm going to debug that before merging this.
Last updated: Nov 22 2024 at 16:03 UTC