cfallin labeled issue #1362:
- It assumes only one new
ValueLabelintroduced at eachSourceLoc. (build_value_labels_index)- It assumes
SourceLocis strictly ascending. (Checking forlast_srcloc.unwrap() > srclocinstead oflast_srcloc.unwrap() != srcloc)- It assumes that a
ValueLabelis only alive for a continuous range ofSourceLoc. (values_labels.range(range))All of those assumptions are broken by cg_clif, meaning that
build_value_labels_rangesoften misses values in the output.
alexcrichton added the wasmtime:debugging label to Issue #1362.
fitzgen closed issue #1362:
- It assumes only one new
ValueLabelintroduced at eachSourceLoc. (build_value_labels_index)- It assumes
SourceLocis strictly ascending. (Checking forlast_srcloc.unwrap() > srclocinstead oflast_srcloc.unwrap() != srcloc)- It assumes that a
ValueLabelis only alive for a continuous range ofSourceLoc. (values_labels.range(range))All of those assumptions are broken by cg_clif, meaning that
build_value_labels_rangesoften misses values in the output.
fitzgen commented on issue #1362:
This does not apply to the new-since-this-issue-was-created mach backends and regalloc2's production of this metadata.
Last updated: Oct 11 2026 at 02:20 UTC