- Refactor caching logic into a helper class PTHSpellingSearch
- Allow "random accesses" in the spelling cache, thus catching the remaining
  cases where 'getSpelling' wasn't hitting the PTH cache
  
For -Eonly, PTH, Cocoa.h:
- This reduces wall time by 3% (user time unchanged, sys time reduced)
- This reduces the amount of paged source by 1112K.
  The remaining 1112K still being paged in is from somewhere else
  (investigating).

llvm-svn: 62009

llvm-svn: 62008

in designated protocols lazily.

llvm-svn: 62007

llvm-svn: 62006

llvm-svn: 62005

llvm-svn: 62004

llvm-svn: 62003

comprehensive and can support more complex style analysis.

llvm-svn: 62002

of ScopedDecls (using the new ScopedDecl::NextDeclInScope
pointer). Performance-wise:

  - It's a net win in memory utilization, since DeclContext is now one
    pointer smaller than it used to be (std::vectors are typically 3
    pointers; we now use 2 pointers) and 
  - Parsing Cocoa.h with -fsyntax-only (with a Release-Asserts Clang)
    is about 1.9% faster than before, most likely because we no longer
    have the memory allocations and copying associated with the
    std::vector.

I'll re-enable serialization of DeclContexts once I've sorted out the
NextDeclarator/NextDeclInScope question.

llvm-svn: 62001

llvm-svn: 62000

llvm-svn: 61999

llvm-svn: 61998

llvm-svn: 61997

Always print out SourceManager stats with 'Stats' is true.  This revealed that PTH always pulls in the source pages with -fsyntax-only (investigating further).

llvm-svn: 61996

llvm-svn: 61995

filters the decls seen by decl_iterator with two criteria: the dynamic
type of the declaration and a run-time predicate described by a member
function. This simplifies EnumDecl, RecordDecl, and ObjCContainerDecl
considerably. It has no measurable performance impact.

llvm-svn: 61994

llvm-svn: 61993

crashes.

llvm-svn: 61992

llvm-svn: 61991

llvm-svn: 61990

llvm-svn: 61989

Add isa/cast/dyncast support for ObjCContainerDecl.
Renamed classprop_iterator/begin/end to prop_iterator/begin/end (the class prefix was confusing).
More simplifications to Sema::ActOnAtEnd()...
Added/changed some FIXME's as a result of the above work.

llvm-svn: 61988

llvm-svn: 61986

I noticed this in the code compiled for a routine using std::map, which produced
this code:
	%25 = tail call i32 @memcmp(i8* %24, i8* %23, i32 6) nounwind readonly
	%.lobit.i = lshr i32 %25, 31		; <i32> [#uses=1]
	%tmp.i = trunc i32 %.lobit.i to i8		; <i8> [#uses=1]
	%toBool = icmp eq i8 %tmp.i, 0		; <i1> [#uses=1]
	br i1 %toBool, label %bb3, label %bb4
which compiled to:

	call	L_memcmp$stub
	shrl	$31, %eax
	testb	%al, %al
	jne	LBB1_11	## 

with this change, we compile it to:

	call	L_memcmp$stub
	testl	%eax, %eax
	js	LBB1_11

This triggers all the time in common code, with patters like this:

	%169 = and i32 %ply, 1		; <i32> [#uses=1]
	%170 = trunc i32 %169 to i8		; <i8> [#uses=1]
	%toBool = icmp ne i8 %170, 0		; <i1> [#uses=1]

 	%7 = lshr i32 %6, 24		; <i32> [#uses=1]
	%9 = trunc i32 %7 to i8		; <i8> [#uses=1]
	%10 = icmp ne i8 %9, 0		; <i1> [#uses=1]

etc

llvm-svn: 61985

llvm-svn: 61984

jump threading can have bugs, who knew? ;-)

llvm-svn: 61983

llvm-svn: 61982

llvm-svn: 61981

llvm-svn: 61980

just llc.

llvm-svn: 61979

llvm-svn: 61978

llvm-svn: 61977

(which is constant time and cheap) before checking hasAllZeroIndices.

llvm-svn: 61976

llvm-svn: 61975

matches llvm-gcc (?).

llvm-svn: 61974

llvm-svn: 61973

llvm-svn: 61972

the same formatting as their corresponding SSE2 instructions, for
consistency.

llvm-svn: 61971

llvm-svn: 61970

llvm-svn: 61969