Account for whether artificial subclasses are skipped on type caching #68404
Conversation
|
@swift-ci smoke test |
Actually in LLDB there is at least one place where we want to not skip artificial subclasses, since we iterate over a type's superclasses ourselves, and if metadata reader silently skipped over them our results would be off. I do agree that this should've probably be defaulted to true instead of false though. |
|
@swift-ci smoke test |
MetadataReader caches types only with the metadata address as the key. However a type lookup can be requested skipping artificial subclasses or not. This makes the cached results incorrect if two requests for the same type, but skipping subclasses on one and not on the other, are made. Fix this by adding a second dimension to the cache key. rdar://101519300
augusto2112
force-pushed
the
type-cache-artificial
branch
from
64f4854 to
9151305
Compare
|
@swift-ci smoke test |
| /// A cache of built types, keyed by the address of the type and whether the | ||
| /// request ignored articial superclasses or not. | ||
| std::unordered_map<std::pair<StoredPointer, bool>, BuiltType, | ||
| TypeCacheKeyHash> |
There was a problem hiding this comment.
Just curious: If we are using the LLVM libraries anyway, why not use a DenseMap?
There was a problem hiding this comment.
Looks like we can, I'll change this in a separate patch.
|
@augusto2112 You will probably also want to cherry-pick this to |
MetadataReader caches types only with the metadata address as the key. However a type lookup can be requested skipping artificial subclasses or not. This makes the cached results incorrect if two requests for the same type, but skipping subclasses on one and not on the other, are made. Fix this by adding a second dimension to the cache key.
rdar://101519300