RobinTD opened Issue #1505:
Using cranelift 0.62, the following results in this error message:" v5 is a ghost value used by a real [Op1ret#c3] instruction".
function u0:82(i64) -> i64 system_v { block0(v0: i64): v1 = load.i64 notrap aligned v0 v2 = load.i64 notrap aligned v0+8 v3 = iconst.i64 1 v4 = iconst.i64 2 v5 = imax v3, v4 return v5 }
RobinTD labeled Issue #1505:
Using cranelift 0.62, the following results in this error message:" v5 is a ghost value used by a real [Op1ret#c3] instruction".
function u0:82(i64) -> i64 system_v { block0(v0: i64): v1 = load.i64 notrap aligned v0 v2 = load.i64 notrap aligned v0+8 v3 = iconst.i64 1 v4 = iconst.i64 2 v5 = imax v3, v4 return v5 }
RobinTD labeled Issue #1505:
Using cranelift 0.62, the following results in this error message:" v5 is a ghost value used by a real [Op1ret#c3] instruction".
function u0:82(i64) -> i64 system_v { block0(v0: i64): v1 = load.i64 notrap aligned v0 v2 = load.i64 notrap aligned v0+8 v3 = iconst.i64 1 v4 = iconst.i64 2 v5 = imax v3, v4 return v5 }
github-actions[bot] commented on Issue #1505:
Subscribe to Label Action
cc @bnjbvr
<details>
This issue or pull request has been labeled: "cranelift"Thus the following users have been cc'd because of the following labels:
- bnjbvr: cranelift
To subscribe or unsubscribe from this label, edit the <code>.github/subscribe-to-label.json</code> configuration file.
Learn more.
</details>
bjorn3 commented on Issue #1505:
imax
is only implemented for 128bit vectors with 8, 16 or 32bit lanes:It makes sense to me to implement it for non-vectors too.
v5 is a ghost value used by a real [Op1ret#c3] instruction
A ghost value is a value which is the output of an instruction without encoding, so this says that the
imax
which hasv5
as output doesn't have an encoding.
RobinTD commented on Issue #1505:
Thanks.
fmin
/fmax
works for scalars, so making it work for integer scalars too would be less surprising.
Last updated: Dec 23 2024 at 12:05 UTC