Fix pairs of doubles using an illegal <8 x i8> vector. #41206
Merged
+40
−9
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Accidentally introduced in #40658 and discovered in some Objective-C bindings (returning
NSPoint
).Turns out LLVM will widen element types of illegal vectors instead of increasing element count, i.e. it will zero-extend
<8 x i8>
to<8 x i16>
, interleaving the bytes, instead of using the first 8 of<16 x i8>
.