Sema: Fix a fuzzer crash #85016
Sema: Fix a fuzzer crash #85016
Conversation
slavapestov force-pushed the prepared-fuzz branch 2 times, most recently from abbd56a to cb58dd8 Compare 
This test case crashes when prepared overloads are disabled, but passes when enabled. To avoid messing up tests if we have to turn the flag on and off, fix the crash.
slavapestov force-pushed the prepared-fuzz branch from cb58dd8 to eaa38f3 Compare 
| @swift-ci Please smoke test |
| for (auto &fix : contextualFixes) | ||
| genericParamInferredTypes.insert(fix.first->getFixedType(resultTypeVar)); | ||
| for (auto &fix : contextualFixes) { | ||
| if (fix.first->hasFixedType(resultTypeVar)) |
There was a problem hiding this comment.
Free type variables are allowed only with FreeTypeVariableBinding::Allow set, I think a better fix here might be to actually get an overload from every solution instead and extra their result type because if hasFixedType is ever produces true it could mean that overloads were chosen in sub-scope when solutions diverged.
There was a problem hiding this comment.
Looks like this is indeed the case:
Solution 1: locator@0xbb4859000 [OverloadedDeclRef@<stdin>:4:12] with Swift.(file).?? as ??: (consuming $T8?, @autoclosure () throws -> $T8) throws -> $T8
Solution 2: locator@0xbb4859000 [OverloadedDeclRef@<stdin>:4:12] with Swift.(file).?? as ??: (consuming $T17?, @autoclosure () throws -> $T17) throws -> $T17
There was a problem hiding this comment.
Ok! I’ll hold off on landing this until I figure out the right fix.
There was a problem hiding this comment.
Slightly higher up is where we get an overload from the first fix, I think all you need to do here is just that for each of the solutions involved.
There was a problem hiding this comment.
extra their result type
This should say "extract" right?
2 participants
This test case crashes when prepared overloads are disabled, but passes when enabled. To avoid messing up tests if we have to turn the flag on and off, fix the crash.