in selectiondag patterns.  This is required for the upcoming shuffle_vector rewrite,
and as it turns out, cleans up a hack in the Alpha instruction info.

llvm-svn: 67286

llvm-svn: 67280

order.

llvm-svn: 67244

  static const WarningOption OptionTable[] = {
    {"unused-macros", DIAGS(UnusedMacrosDiags)}
    ...
  };

This table is not yet properly sorted.

llvm-svn: 67242

declarations for controlling groups of warnings. Currently this transforms:

  def UnusedMacrosDiags : Option<"unused-macros", [pp_macro_not_used]>;
  
into:

  static const diag::kind UnusedMacrosDiags[] = { diag::pp_macro_not_used };

llvm-svn: 67239

This allows tblgen to handle include "foo.td" when the quote is exactly
the last character in a file.  rdar://6695728

llvm-svn: 67232

llvm-svn: 67215

for pointing this out :)

llvm-svn: 67212

llvm-svn: 67211

llvm-svn: 67210

unused loads or bitcasts.

llvm-svn: 67202

If the instruction has no users, it is also not only used by debug info 
and should not be deleted.

llvm-svn: 67194

both in Linux and Darwin.

llvm-svn: 67191

If I->use_empty(), this method should return false.

llvm-svn: 67180

and expanding a bit convert (PR3711).  In both cases, we extract the
valid part of the widen vector and then do the conversion.

llvm-svn: 67175

llvm-svn: 67156

llvm-svn: 67143

not safe in general because the immediate could be an arbitrary
value that does not fit in a 32-bit pcrel displacement.  
Conservatively fall back to loading the value into a register
and calling through it.

We still do the optzn on X86-32.

llvm-svn: 67142

an invoke instead of after the invoke (in its block), which is invalid.

llvm-svn: 67139

llvm-svn: 67138

llvm-svn: 67137

it is not APInt clean, but even when it is it needs to be evaluated carefully
to determine whether it is actually profitable.

This fixes a crash on PR3806

llvm-svn: 67134

Some architectures (like x86) don't require it.
This fixes bug 3779.

llvm-svn: 67132

 - Use for exceptional buffer conditions in raw_ostream:write to shave
   off a cycle or two.

 - Please rename if you have a better one.

llvm-svn: 67103

to see if this is producing the expected code or not, I'm
not sure what the test was intended to check.

llvm-svn: 67099

llvm-svn: 67096

size by the array amount as an i32 value instead of promoting from
i32 to i64 then doing the multiply.  Not doing this broke wrap-around
assumptions that the optimizers (validly) made.  The ultimate real
fix for this is to introduce i64 version of alloca and remove mallocinst.

This fixes PR3829

llvm-svn: 67093

llvm-svn: 67089

llvm-svn: 67085

Revert inadvertent mis-fix of fneg.

llvm-svn: 67084

llvm-svn: 67083

llvm-svn: 67082

llvm-svn: 67080

(which produces "call L_f$stub" rather than "call f").

llvm-svn: 67079

vector shuffle mask. Forced the mask to be built using i32.  Note: this will
be irrelevant once vector_shuffle no longer takes a build vector for the
shuffle mask.

llvm-svn: 67076

llvm-svn: 67072

llvm-svn: 67071

write) to expose more alias information.

llvm-svn: 67070

under a single branch.

Also, add a FIXME for formatted output.

llvm-svn: 67069

Spiller may unfold load / mod / store instructions as an optimization when the would be loaded value is available in a register. It needs to check if it's legal to clobber the register. Also, the register can contain values of multiple spill slots, make sure to check all instead of just the one being unfolded.

llvm-svn: 67068