To import a type identifier we read the summary and create external
references to the symbols defined when exporting.
Details
Diff Detail
- Build Status
Buildable 3128 Build 3128: arc lint + arc unit
Event Timeline
llvm/lib/Transforms/IPO/LowerTypeTests.cpp | ||
---|---|---|
743 | This function does not import SizeM1BitWidth. | |
754 | I don't get this condition. Why would this variable exist and be non-hidden? | |
766 | should this say 0ull, ~0ull ? |
llvm/lib/Transforms/IPO/LowerTypeTests.cpp | ||
---|---|---|
743 | No, we don't need it. Removed in r292647. | |
754 | If getOrInsertGlobal created the global, it would have default visibility. Added a comment. | |
766 | "~0ull, ~0ull" means the full set, whereas "0ull, ~0ull" means the full set except for the value ~0ull. From http://llvm.org/docs/LangRef.html#range-metadata : "The pair a,b represents the range [a,b).". See also http://llvm-cs.pcc.me.uk/lib/IR/ConstantRange.cpp#237 Added a comment. |
llvm/lib/Transforms/IPO/LowerTypeTests.cpp | ||
---|---|---|
766 | From http://llvm.org/docs/LangRef.html#range-metadata, The range should not represent the full or empty set. That is, a!=b. Does the documentation need to be updated? |
llvm/lib/Transforms/IPO/LowerTypeTests.cpp | ||
---|---|---|
766 | The doc for range is correct, but the one for absolute_symbol needs to be corrected. I'll do that. |
This function does not import SizeM1BitWidth.
I realize it is unused, but this could be a surprise for someone later, comments in TypeIdLowering say that it should be set for ByteArray, Inline, AllOnes.
Do we even need this field? It looks like it could be reconstructed from SizeM1.