llvm-svn: 90180

llvm-svn: 90176

ASTContext instead of malloc.  Besides reducing malloc traffic, this
also removes a source of memory leaks when using a BumpPtrAllocator
for the allocator of ASTContext.  There are still leaks when using
MallocAllocator because Decl::Destroy() isn't fully finished.

Fixes: <rdar://problem/7431556>
llvm-svn: 90174

For VMOVv*i[16,32], op bit is don't care, and some cmode bits vary depending on
the immediate values.

Ref: Table A7-15 Modified immediate values for Advanced SIMD instructions.
llvm-svn: 90173

llvm-svn: 90172

IDs in dependent contexts are not dependent if the context names a namespace.

llvm-svn: 90171

llvm-svn: 90170

llvm-svn: 90169

Have ASTRecordLayout keep track of the key function, in preparation of fixing a synthetic ctor/dtor bug.

llvm-svn: 90168

llvm-svn: 90166

llvm-svn: 90165

llvm-svn: 90164

New virtual registers created for spill intervals should inherit allocation hints from the original register.

This helps us avoid silly copies when rematting values that are copied to a physical register:

leaq	_.str44(%rip), %rcx
movq	%rcx, %rsi
call	_strcmp

becomes:

leaq	_.str44(%rip), %rsi
call	_strcmp

The coalescer will not touch the movq because that would tie down the physical register.

llvm-svn: 90163

those associated with TemplateNames.

llvm-svn: 90162

Create a new UnresolvedMemberExpr for these lookups.  Assorted hackery
around qualified member expressions;  this will all go away when we
implement the correct (i.e. extremely delayed) implicit-member semantics.

llvm-svn: 90161

llvm-svn: 90160

llvm-svn: 90159

Fixes pr5619

llvm-svn: 90158

When we're trying to define an implicit virtual destructor, make sure that we have a valid delete operator.

llvm-svn: 90156

llvm-svn: 90155

llvm-svn: 90154

llvm-svn: 90153

llvm-svn: 90152

loops (including temporaries inside the initializers).

llvm-svn: 90151

llvm-svn: 90149

branches even when optimizing for code size.  Unless we find evidence to the
contrary in the future, the special treatment for indirect branches does not
have a significant effect on code size, and performance still matters with -Os.

llvm-svn: 90147

recent change to make SEL a builtin type (fixes radar 7425510).

llvm-svn: 90145

for all the processors where I have tried it, and even when it might not help
performance, the cost is quite low.  The opportunities for duplicating
indirect branches are limited by other factors so code size does not change
much due to tail duplicating indirect branches aggressively.

llvm-svn: 90144

feel free to extend!

llvm-svn: 90143

llvm-svn: 90141

llvm-svn: 90140

llvm-svn: 90139

llvm-svn: 90138

llvm-svn: 90137

llvm-svn: 90136

llvm-svn: 90135

llvm-svn: 90134

llvm-svn: 90133

If no destination label is available, just point to the node itself
instead of pointing to some source label. Source and destination labels are
not related in any way.

llvm-svn: 90132

Graphviz can layout the graphs better if a node does not contain source
ports. Therefore only print the ports if the source ports are useful,
that means are not labeled with the empty string "".
This patch also simplifies graphs without any edgeSourceLabels e.g. the
dominance trees.

llvm-svn: 90131