- Overhauled the notion of "types" for TypedRegions. We now distinguish between the "lvalue" of a region (via getLValueRegion()) and the "rvalue" of a region (va getRValueRegion()). Since a region represents a chunk of memory it has both, but we were conflating these concepts in some cases, leading to some insidious bugs.
- Removed AnonPointeeType, partially because it is unused and because it doesn't have a clear notion of lvalue vs rvalue type. We can add it back once there is a need for it and we can resolve its role with these concepts.
StoreManager:
- Overhauled StoreManager::CastRegion. It expects an *lvalue* type for a region. This is actually what motivated the overhaul to the MemRegion type mechanism. It also no longer returns an SVal; we can just return a MemRegion*.
- BasicStoreManager::CastRegion now overlays an "AnonTypedRegion" for pointer-pointer casts. This matches with the MemRegion changes.
- Similar changes to RegionStore, except I've added a bunch of FIXMEs where it wasn't 100% clear where we should use TypedRegion::getRValueRegion() or TypedRegion::getLValueRegion().
AuditCFNumberCreate check:
- Now blasts through AnonTypedRegions that may layer the original memory region, thus checking if the actually memory block is of the appropriate type. This change was needed to work with the changes to StoreManager::CastRegion.
GRExprEngine::VisitCast:
- Conform to the new interface of StoreManager::CastRegion.
Tests:
- None of the analysis tests fail now for using the "basic store".
- Disabled the tests 'array-struct.c' and 'rdar-6442306-1.m' pending further testing and bug fixing.
llvm-svn: 60995
GRExprEngine (VisitCast):
- When using StoreManager::CastRegion, always use the state and value it returns to generate the next node. Failure to do so means that region values returned that don't require the state to be modified will get ignored.
MemRegion:
- Tighten the interface for ElementRegion. Now ElementRegion can only be created with a super region that is a 'TypedRegion' instead of any MemRegion. Code in BasicStoreManager/RegionStoreManager already assumed this, but it would result in a dynamic assertion check (and crash) rather than just having the compiler forbid the construction of such regions.
- Added ElementRegion::getArrayRegion() to return the 'typed version' of an ElementRegion's super region.
- Removed bogus assertion in ElementRegion::getType() that assumed that the super region was an AnonTypedRegion. All that matters is that it is a TypedRegion, which is now true all the time by design.
BasicStore:
- Modified getLValueElement() to check if the 'array' region is a TypedRegion before creating an ElementRegion. This conforms to the updated interface for ElementRegion.
RegionStore:
- In ArrayToPointer() gracefully handle things we don't reason about, and only create an ElementRegion if the array region is indeed a TypedRegion.
llvm-svn: 60990
llvm[2]: Linking Release executable opt (without symbols)
...
Undefined symbols:
"llvm::APFloat::IEEEsingle", referenced from:
__ZN4llvm7APFloat10IEEEsingleE$non_lazy_ptr in libLLVMCore.a(Constants.o)
__ZN4llvm7APFloat10IEEEsingleE$non_lazy_ptr in libLLVMCore.a(AsmWriter.o)
__ZN4llvm7APFloat10IEEEsingleE$non_lazy_ptr in libLLVMCore.a(ConstantFold.o)
"llvm::APFloat::IEEEdouble", referenced from:
__ZN4llvm7APFloat10IEEEdoubleE$non_lazy_ptr in libLLVMCore.a(Constants.o)
__ZN4llvm7APFloat10IEEEdoubleE$non_lazy_ptr in libLLVMCore.a(AsmWriter.o)
__ZN4llvm7APFloat10IEEEdoubleE$non_lazy_ptr in libLLVMCore.a(ConstantFold.o)
ld: symbol(s) not found
This is in release mode. To replicate, compile llvm and llvm-gcc in optimized
mode. Then build llvm, in optimized mode, with the newly created compiler.
llvm-svn: 60977
width register load followed by a truncating
store for the copy, since the load will not place
the value in the lower bits. Probably partial
loads/stores can never happen here, but fix it
anyway.
llvm-svn: 60972
use of illegal integer types: instead, use a stack slot
and copying via integer registers. The existing code
is still used if the bitconvert is to a legal integer
type.
This fires on the PPC testcases 2007-09-08-unaligned.ll
and vec_misaligned.ll. It looks like equivalent code
is generated with these changes, just permuted, but
it's hard to tell.
With these changes, nothing in LegalizeDAG produces
illegal integer types anymore. This is a prerequisite
for removing the LegalizeDAG type legalization code.
While there I noticed that the existing code doesn't
handle trunc store of f64 to f32: it turns this into
an i64 store, which represents a 4 byte stack smash.
I added a FIXME about this. Hopefully someone more
motivated than I am will take care of it.
llvm-svn: 60964
which are identical to the original patterns.
- Change the multiply with overflow so that we distinguish between signed and
unsigned multiplication. Currently, unsigned multiplication with overflow
isn't working!
llvm-svn: 60963
Misha Brukman