keep f32 in SSE registers and f64 in x87.  This
is effectively a new codegen mode.
Change addLegalFPImmediate to permit float and
double variants to do different things.
Adjust callers.

llvm-svn: 42246

llvm-svn: 42239

llvm-svn: 42222

llvm-svn: 42220

bit width instead of number of words allocated, which
makes it actually work for int->APF conversions.
Adjust callers.  Add const to one of the APInt constructors
to prevent surprising match when called with const
argument.

llvm-svn: 42210

llvm-svn: 42209

llvm-svn: 42205

llvm-svn: 42204

llvm-svn: 42193

Implement calls to functions with byval arguments on X86

llvm-svn: 42192

llvm-svn: 42189

speedup for GVN.

llvm-svn: 42185

llvm-svn: 42182

llvm-svn: 42181

llvm-svn: 42180

llvm-svn: 42179

llvm-svn: 42178

llvm-svn: 42165

llvm-svn: 42162

llvm-svn: 42159

llvm-svn: 42157

llvm-svn: 42156

llvm-svn: 42155

llvm-svn: 42153

llvm-svn: 42149

llvm-svn: 42147

llvm-svn: 42143

llvm-svn: 42140

help non-local memdep caching.

llvm-svn: 42137

llvm-svn: 42133

llvm-svn: 42132

fold that were missed in the fix for PR1646.  Probably
this null/not-null logic should be factorized somewhere.

llvm-svn: 42131

an alias could alias such a global variable.

llvm-svn: 42130

in tracking physical register output dependencies.

llvm-svn: 42125

llvm-svn: 42124

between two registers in the specific class.

llvm-svn: 42123

llvm-svn: 42122

for (int i = 0; i < N; ++i) {
  if (i == somevalue)
    dosomething();
   else
    dosomethingelse();
}

llvm-svn: 42121

llvm-svn: 42120

llvm-svn: 42119