Let each target determines whether a machine instruction is dead. If true, that allows late codeine passes to delete it.

This is considered a workaround. The problem is some targets are not modeling side effects correctly. PPC is apparently one of those. This patch allows ppc llvm-gcc to bootstrap on Darwin. Once we find out which instruction definitions are wrong, we can remove the PPCInstrInfo workaround.

llvm-svn: 76703

llvm-svn: 76702

Constant::getRelocationInfo(), which has a much simpler
to use API.  It still should not be part of libvmcore, but
is better than it was.  Also teach it to be smart about 
hidden visibility.

llvm-svn: 76700

llvm-svn: 76699

negative on an individual bases rather than basing on whether it's in thumb
mode.

llvm-svn: 76698

llvm-svn: 76693

member functions of class templates, e.g.,

  template<typename T> 
  struct X {
    void f(T);
  };

  template<typename T> X<T>::f(T) { /* ... */ }

llvm-svn: 76692

llvm-svn: 76691

llvm-svn: 76690

llvm-svn: 76689

llvm-svn: 76688

llvm-svn: 76687

virtual.  Just inline it into its two current call sites in preparation
for simplifying the code.

llvm-svn: 76686

llvm-svn: 76685

pinsrd in sse4.1.

llvm-svn: 76684

llvm-svn: 76683

llvm-svn: 76682

llvm-svn: 76681

Add some support for iterative coalescers to calculate a joined live
range's weight properly.  This is turned off right now in the sense that
you'll get an assert if you get into a situation that can only be caused
by an iterative coalescer.  All other code paths operate exactly as
before so there is no functional change with this patch.  The asserts
should be disabled if/when an iterative coalescer gets added to trunk.

llvm-svn: 76680

llvm-svn: 76679

pool entry will require relocations against it.  I implemented this
conservatively for ARM, someone who is knowledgable about it should
see if this can be improved.

llvm-svn: 76678

poisoning changes to DenseMap. We were using an iterator after it had been
invalidated by an insertion into the DenseMap.

llvm-svn: 76677

getAnalysisIfAvailable<TargetData>.

llvm-svn: 76676

to help support use when TargetData is not available.

llvm-svn: 76675

llvm-svn: 76674

their appropriate sections before the code itself. They need to be emitted
before the function because on some targets (x86 but not x86_64) the later
may reference a JT or CP entry address

llvm-svn: 76672

(x pred y) with more thorough code that does more complete canonicalization
before resorting to range checks. This helps it find more cases where
the canonicalized expressions match.

llvm-svn: 76671

llvm-svn: 76670

llvm-svn: 76669

llvm-svn: 76668

depends on XS1A, but I think the ReadOnlySection is already set up for this
and there is no testcase that this breaks.  If this is really needed, we can
add the appropriate parameterization to TargetAsmInfo in the future to support
this.

llvm-svn: 76667

llvm-svn: 76666

if Xcore doesn't support TLS, it doesn't have to worry about thread local LLVM IR, it should be rejected by a front-end.

llvm-svn: 76665

to twist your brain to see it, I believe it is the same as ELFTargetAsmInfo::SelectSectionForGlobal.

llvm-svn: 76664

destruction of base/members for each destructor AST.

llvm-svn: 76663

llvm-svn: 76662

llvm-svn: 76661

llvm-svn: 76660

the generic ELF version instead.  This will result in its mergable constant
sections getting named ".rodata.cst4" instead of ".cp.const4", but the
linker looks at the section flags, not the name of the section AFAICT.

llvm-svn: 76659

as 32-bit code by default, and if gcc defaults to 64-bit code then ocamlc
requires a -cc "gcc -arch i386" option.  We were hardcoding -cc g++
and throwing away any other compiler options that were determined when
ocamlc was configured and built.

llvm-svn: 76658