Add semi-dynamic ccall support
#69
Comments
|
This is not a soundness bug, but it does mean that some dynamic ccalls that we could statically resolve to the runtime are instead treated as DynamicDispatch |
topolarity
added a commit
that referenced
this issue
Oct 8, 2024
These can occasionally happen due to known incomplete spots in our call resolution (#69) or due to changes in the optimizer, etc. Rather than conservatively calling these "dynamic dispatches", this adds a new error type that encourages the user to file a bug upstream, since a feature-complete AllocCheck should never encounter them.
topolarity
added a commit
that referenced
this issue
Oct 8, 2024
These can occasionally happen due to known incomplete spots in our call resolution (#69) or due to changes in the optimizer, etc. Rather than conservatively calling these "dynamic dispatches", this adds a new error type that encourages the user to file a bug upstream, since a feature-complete AllocCheck should never encounter them.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
We need to expand our resolution here to walk through the
PhiNodein the LLVM IR and notice the ccall /jl_load_and_lookupmachinery.Originally posted by @topolarity in #68 (comment)
The text was updated successfully, but these errors were encountered: