Chandler Carruth 9fc5a53118 [PM] Fix a pretty nasty bug where the new pass manager would invalidate ... passes too many time. I think this is actually the issue that someone raised with me at the developer's meeting and in an email, but that we never really got to the bottom of. Having all the testing utilities made it much easier to dig down and uncover the core issue. When a pass manager is running many passes over a single function, we need it to invalidate the analyses between each run so that they can be re-computed as needed. We also need to track the intersection of preserved higher-level analyses across all the passes that we run (for example, if there is one module analysis which all the function analyses preserve, we want to track that and propagate it). Unfortunately, this interacted poorly with any enclosing pass adaptor between two IR units. It would see the intersection of preserved analyses, and need to invalidate any other analyses, but some of the un-preserved analyses might have already been invalidated *and recomputed*! We would fail to propagate the fact that the analysis had already been invalidated. The solution to this struck me as really strange at first, but the more I thought about it, the more natural it seemed. After a nice discussion with Duncan about it on IRC, it seemed even nicer. The idea is that invalidating an analysis *causes* it to be preserved! Preserving the lack of result is trivial. If it is recomputed, great. Until something *else* invalidates it again, we're good. The consequence of this is that the invalidate methods on the analysis manager which operate over many passes now consume their PreservedAnalyses object, update it to "preserve" every analysis pass to which it delivers an invalidation (regardless of whether the pass chooses to be removed, or handles the invalidation itself by updating itself). Then we return this augmented set from the invalidate routine, letting the pass manager take the result and use the intersection of *that* across each pass run to compute the final preserved set. This accounts for all the places where the early invalidation of an analysis has already "preserved" it for a future run. I've beefed up the testing and adjusted the assertions to show that we no longer repeatedly invalidate or compute the analyses across nested pass managers. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@225333 91177308-0d34-0410-b5e6-96231b3b80d8		2015-01-07 01:58:35 +00:00
..
AsmWriter.cpp	Change the .ll syntax for comdats and add a syntactic sugar.	2015-01-06 22:55:16 +00:00
AsmWriter.h
AttributeImpl.h
Attributes.cpp
AutoUpgrade.cpp	IR: Split Metadata from Value	2014-12-09 18:38:53 +00:00
BasicBlock.cpp	The leak detector is dead, long live asan and valgrind.	2014-12-22 13:00:36 +00:00
CMakeLists.txt	The leak detector is dead, long live asan and valgrind.	2014-12-22 13:00:36 +00:00
Comdat.cpp
ConstantFold.cpp	ConstantFold: Shifting undef by zero results in undef	2014-12-18 23:54:43 +00:00
ConstantFold.h
ConstantRange.cpp
Constants.cpp
ConstantsContext.h
Core.cpp	IR: Split Metadata from Value	2014-12-09 18:38:53 +00:00
DataLayout.cpp	DataLayout: Provide nicer diagnostics for malformed strings	2014-12-10 02:36:41 +00:00
DebugInfo.cpp	Remove isSubroutineType test for isCompositeType, getTag() is enough.	2014-12-19 22:15:09 +00:00
DebugLoc.cpp	IR: Split Metadata from Value	2014-12-09 18:38:53 +00:00
DiagnosticInfo.cpp	IR: Split Metadata from Value	2014-12-09 18:38:53 +00:00
DiagnosticPrinter.cpp
DIBuilder.cpp	DIBuilder: Similar to createPointerType, make createMemberPointerType take	2014-12-23 19:11:47 +00:00
Dominators.cpp
Function.cpp	Masked Load/Store - Changed the order of parameters in intrinsics.	2014-12-25 07:49:20 +00:00
GCOV.cpp
Globals.cpp	The leak detector is dead, long live asan and valgrind.	2014-12-22 13:00:36 +00:00
GVMaterializer.cpp
InlineAsm.cpp
Instruction.cpp	The leak detector is dead, long live asan and valgrind.	2014-12-22 13:00:36 +00:00
Instructions.cpp	IR: Split Metadata from Value	2014-12-09 18:38:53 +00:00
IntrinsicInst.cpp	IR: Split Metadata from Value	2014-12-09 18:38:53 +00:00
IRBuilder.cpp	Some code improvements in Masked Load/Store.	2014-12-30 14:28:14 +00:00
IRPrintingPasses.cpp	[PM] Switch the new pass manager to use a reference-based API for IR	2015-01-05 02:47:05 +00:00
LeaksContext.h	IR: Store MDNodes in a separate LeakDetector container	2014-12-11 21:39:39 +00:00
LegacyPassManager.cpp	Document that PassManager::add() may delete the pass right away.	2014-12-12 01:27:01 +00:00
LLVMBuild.txt
LLVMContext.cpp	Fix LLVMContext to match what MDKind names that the LL parser permits. Fixes PR21799!	2014-12-11 02:10:28 +00:00
LLVMContextImpl.cpp	IR: Split Metadata from Value	2014-12-09 18:38:53 +00:00
LLVMContextImpl.h	IR: Split Metadata from Value	2014-12-09 18:38:53 +00:00
Makefile
Mangler.cpp
MDBuilder.cpp	IR: Split Metadata from Value	2014-12-09 18:38:53 +00:00
Metadata.cpp	IR: Don't drop MDNode uniquing on null operands	2015-01-05 23:31:54 +00:00
MetadataTracking.cpp	IR: Split Metadata from Value	2014-12-09 18:38:53 +00:00
Module.cpp	The leak detector is dead, long live asan and valgrind.	2014-12-22 13:00:36 +00:00
module.modulemap
Pass.cpp
PassManager.cpp	[PM] Fix a pretty nasty bug where the new pass manager would invalidate	2015-01-07 01:58:35 +00:00
PassRegistry.cpp
Statepoint.cpp
SymbolTableListTraitsImpl.h
Type.cpp
TypeFinder.cpp	IR: Split Metadata from Value	2014-12-09 18:38:53 +00:00
Use.cpp
UseListOrder.cpp
User.cpp
Value.cpp	The leak detector is dead, long live asan and valgrind.	2014-12-22 13:00:36 +00:00
ValueSymbolTable.cpp	IR: Split Metadata from Value	2014-12-09 18:38:53 +00:00
ValueTypes.cpp
Verifier.cpp	[PM] Switch the new pass manager to use a reference-based API for IR	2015-01-05 02:47:05 +00:00