llvm-svn: 98161

llvm-svn: 98160

Really apply (unnoticed weird git-svn merge conflict in 98144): Refactor RegionStore::RemoveDeadBindings to use the same core cluster analysis algorithm as RegionStore::InvalidateRegions().

Beyond simplifying the algorithm significantly, we no longer
need to build subregion maps in RemoveDeadBindings().  This
and other changes cause a significant speedup: the time to
analyze sqlite3.c (single core) drops by 14%.

llvm-svn: 98159

llvm-svn: 98158

results are unused elsewhere.

llvm-svn: 98157

llvm-svn: 98156

I'll get this loop right yet.

llvm-svn: 98155

llvm-svn: 98154

llvm-svn: 98153

llvm-svn: 98152

llvm-svn: 98151

directly to the maccu / maccs instructions. We handle this in
ExpandADDSUB since after type legalisation it is messy to
recognise these operations.

llvm-svn: 98150

therefore not creating ElaboratedTypes, which are still pretty-printed
with the written tag).

Most of these testcase changes were done by script, so don't feel too
sorry for my fingers.

llvm-svn: 98149

llvm-svn: 98148

semantic instead of syntactic.  This completes MCization of
darwin/x86[-64]!

llvm-svn: 98145

cluster analysis algorithm as RegionStore::InvalidateRegions().

Beyond simplifying the algorithm significantly, we no longer
need to build subregion maps in RemoveDeadBindings().  This
and other changes cause a significant speedup: the time to
analyze sqlite3.c (single core) drops by 14%.

llvm-svn: 98144

into a 'ClusterAnalysis' parent class.  The idea is
to potentially reuse this for reworking RemoveDeadBindings.

llvm-svn: 98143

llvm-svn: 98142

llvm-svn: 98141

llvm-svn: 98140

Don't accidentally mark some functions in construction vtables as unused. Also land the test for a previous checkin, now that it's correct.

llvm-svn: 98139

llvm-svn: 98138

llvm-svn: 98137

llvm-svn: 98136

llvm-svn: 98135

injected class name of a class template or class template partial specialization.
This is a non-canonical type;  the canonical type is still a template 
specialization type.  This becomes the TypeForDecl of the pattern declaration,
which cleans up some amount of code (and complicates some other parts, but
whatever).

Fixes PR6326 and probably a few others, primarily by re-establishing a few
invariants about TypeLoc sizes.     

llvm-svn: 98134

llvm-svn: 98132

When building construction vtables, we need to check if a primary virtual base is actually a primary virtual base in the layout class.

llvm-svn: 98131

for darwin/x86 to be completely mcized.

llvm-svn: 98130

llvm-svn: 98129

Improve vcall offset handling in construction vtables. With this we layout the construction vtables from the ABI examples correctly.

llvm-svn: 98127

llvm-svn: 98125

llvm-svn: 98124

This fixes PR6474.

llvm-svn: 98123

llvm-svn: 98122

1. Be careful with cse "cheap" expressions. e.g. constant materialization. Only cse them when the common expression is local or in a direct predecessor. We don't want cse of cheap instruction causing other expressions to be spilled.
2. Watch out for the case where the expression doesn't itself uses a virtual register. e.g. lea of frame object. If the common expression itself is used by copies (common for passing addresses to function calls), don't perform the cse. Since these expressions do not use a register, it creates a live range but doesn't close any, we want to be very careful with increasing register pressure.

Note these are heuristics so machine cse doesn't make register allocator unhappy. Once we have proper live range splitting and re-materialization support in place, these should be evaluated again.

Now machine cse is almost always a win on llvm nightly tests on x86 and x86_64.

llvm-svn: 98121

llvm-svn: 98120

llvm-svn: 98119

Add a new GetOrCreateTemporarySymbol method and a version that
takes a twine.

llvm-svn: 98118

dead IsPCRel argument.

llvm-svn: 98117