llvm-svn: 14172

is write an autoconf macro that checks whether __isnan or isnan actually works
**using the C++ compiler after #include <cmath>**, instead of doing it the easy
way with AC_CHECK_FUNCS().

llvm-svn: 14171

peepholing, and make it work the same way in the JIT as in LLC.

llvm-svn: 14170

llvm-svn: 14169

llvm-svn: 14168

llvm-svn: 14167

function verification and make it a requirement that both arguments to
llvm.isunordered are of the same type.

llvm-svn: 14165

llvm-svn: 14159

on a 1280x1024 screen!

llvm-svn: 14158

llvm-svn: 14157

llvm-svn: 14156

llvm-svn: 14155

instructions generated.

llvm-svn: 14154

-release causes ENABLE_OPTIMIZED=1 to be added to the make line so that
         the test runs against optimized code.
-pedantic causes the CompileOptimizeOpts variable to be overridden to
          add several additional warnings not picked up by -Wall. This
	  should help catch additional programming faux pas. See the diff
	  for the specific details.

llvm-svn: 14153

printed incorrectly, and we were reserving 8 extra bytes of stack space for functions
on X86.

llvm-svn: 14152

llvm-svn: 14151

llvm-svn: 14150

llvm-svn: 14149

comparisons.  In an 'isunordered' predicate, which looks like this at
the LLVM level:

        %a = call bool %llvm.isnan(double %X)
        %b = call bool %llvm.isnan(double %Y)
        %COM = or bool %a, %b

We used to generate this code:

        fxch %ST(1)
        fucomip %ST(0), %ST(0)
        setp %AL
        fucomip %ST(0), %ST(0)
        setp %AH
        or %AL, %AH

With this patch, we generate this code:

        fucomip %ST(0), %ST(1)
        fstp %ST(0)
        setp %AL

Which should make alkis happy.  Tested as X86/compare_folding.llx:test1

llvm-svn: 14148

for each 'COM =' line.

llvm-svn: 14147

llvm-svn: 14146

llvm-svn: 14145

we can get rid of the FpUCOM/FpUCOMi pseudo instructions, which makes stuff simpler
and faster.

llvm-svn: 14144

twoarg cases.

llvm-svn: 14143

llvm-svn: 14142

testcase

llvm-svn: 14141

llvm-svn: 14140

This makes the code much simpler, and the two cases really do belong apart.
Once we do it, it's pretty obvious how flawed the logic was for A != A case,
so I fixed it (fixing PR369).

This also uses freeStackSlotAfter instead of inserting an fxchg then
popStackAfter'ing in the case where there is a dead result (unlikely, but
possible), producing better code.

llvm-svn: 14139

doesn't break on machines with limited command line lengths.

llvm-svn: 14138

llvm-svn: 14137

llvm-svn: 14136

llvm-svn: 14135

llvm-svn: 14134

  isunordered(X, Y) === isnan(X) | isnan(Y)

Remove isunordered, add isnan.

Modernize testcase

llvm-svn: 14133

  isunordered(X, Y) === isnan(X) | isnan(Y)

Remove isunordered, add isnan.

llvm-svn: 14132

llvm-svn: 14131

llvm-svn: 14130

llvm-svn: 14129

them into the constant pool.

llvm-svn: 14128

llvm-svn: 14127