1. Actually turn on -fno-exceptions in libraries that do not have the
   REQUIRES_EH option in their Makefile. The following library file size
   savings were made (DEBUG):
      libLLVMDataStructure.a  525K
      libLLVMCore.a           380K
      libLLVMCodeGen.a        350K
      libLLVMTransformUtils.a 305K
      libLLVMScalarOpts.a     270K
      libLLVMAnalysis.a       247K
      libLLVMSelectionDAG.a   233K
      libLLVMipo.a            175K
      LLVMX86.o               123K
      LLVMPPC.o                81K
      libLLVMipa.a             17K
                 TOTAL      2,706K
      Note that the savings is actually a little larger than this because
      I didn't count any of the libraries that had small changes.

2. Remove REQUIRES_EH from the AsmParser library as it is now exception
   free. This resulted in a nearly 78K drop in the size of the debug
   library for AsmParser.

llvm-svn: 29767

Update to reflect ParseException becoming ParseError (from Parser.h)

llvm-svn: 29766

llvm-svn: 29765

specification can't be handled by llvmc (= confuses it)

llvm-svn: 29764

Rid the Assembly Parser of exceptions. This is a really gross hack but it
will do until the Assembly Parser is re-written as a recursive descent.
The basic premise is that wherever the old "ThrowException" function was
called (new name: GenerateError) we set a flag (TriggerError). Every
production checks that flag and calls YYERROR if it is set. Additionally,
each call to ThrowException in the grammar is replaced with GEN_ERROR
which calls GenerateError and then YYERROR immediately. This prevents
the remaining production from continuing after an error condition.

llvm-svn: 29763

Shrinkify LLVM's footprint by removing the analyze tool and moving its
functionality into the opt tool. THis eliminates one of the largest tools
from LLVM and doesn't make opt much bigger because it already included
most of the analysis passes.  To get the old analyze functionality pass
the -analyze option to opt. Note that the integeration here is dead
simple. The "main" of analyze was just copied to opt and invoked if the
-analyze option was given. There may be opportunities for further
integration such as removing the distinction between transform passes
and analysis passes.

To use the analysis functionality, if you previously did this:
  analyze $FNAME -domset -disable-verify
you would now do this:
  opt -analyze $FNAME -domset -disable-verify
Pretty simple.

llvm-svn: 29762

llvm-svn: 29760

Changing the format dejagnu tests are submitted. We used to only submit multisource information for pass fail. Now we submit all pass/fail test information. This should fix the issue where some results were not being reported.

llvm-svn: 29757

llvm-svn: 29756

llvm-svn: 29754

llvm-svn: 29753

llvm-svn: 29752

llvm-svn: 29751

llvm-svn: 29750

llvm-svn: 29748

llvm-svn: 29747

llvm-svn: 29746

llvm-svn: 29745

can be added to the faq.

llvm-svn: 29744

This reduces selectiondag time on kc++ from 5.43s to 4.98s (9%).  More
significantly, this speeds up the default ppc scheduler from ~1571ms to 1063ms,
a 33% speedup.

llvm-svn: 29743

llvm-svn: 29742

llvm-svn: 29741

llvm-svn: 29740

llvm-svn: 29739

My last commit added some experimental features I have been working on. This commit rolls those back.

llvm-svn: 29738

llvm-svn: 29736

llvm-svn: 29735

llvm-svn: 29734

llvm-svn: 29732

isel time from 6.79s to 5.53s (22%) on kc++ with a release build on ppc.
Go smallvector! :)

llvm-svn: 29731

llvm-svn: 29730

Patches contributed by Bram Adams. Thanks Bram.

llvm-svn: 29729

Remove the hard coded store and load of the link register
Implement ARMFrameInfo

llvm-svn: 29727

llvm-svn: 29726

llvm-svn: 29725

llvm-svn: 29724

llvm-svn: 29723

llvm-svn: 29722

linking the program on the fly.

llvm-svn: 29721

llvm-svn: 29720