This kind of simplification is sometimes useful, but in general it's not correct. 

As GNU/kFreeBSD is an hybrid system, for kernel-related issues we want to match the
build definitions used for FreeBSD, whereas for userland-related issues we want to
match the definitions used for other systems with Glibc.

The current modification adjusts the build system so that they can be distinguished,
and explicitly adds GNU/kFreeBSD to the build checks in which it belongs.

Fixes bug #16444.

Patch by Robert Millan in the context of Debian.

llvm-svn: 185311

Math functions are mark as readonly because they read the floating point
rounding mode. Because we don't vectorize loops that would contain function
calls that set the rounding mode it is safe to ignore this memory read.

llvm-svn: 185299

llvm-svn: 185294

llvm-svn: 185292

DeadArgumentElimination: keep return value on functions that have a live argument with the 'returned' attribute (rather than generate invalid IR); however, if both can be eliminated, both will be

llvm-svn: 185290

ConstantFold: Check that truncating the other side is safe under a sext when trying to remove a sext from a compare.

Fixes PR16462.

llvm-svn: 185284

This allows us to simplify urem instructions involving the add+xor to
turn into simpler math.

llvm-svn: 185272

Avoids unused variable warnings in release builds.

llvm-svn: 185271

Inserting a zext or trunc is sufficient. This pattern is somewhat common in
LLVM's pointer mangling code.

llvm-svn: 185270

operator++ on an enum is not legal. clang happens to accept it anyways, I think
that's a known bug.

llvm-svn: 185269

llvm-svn: 185268

llvm-svn: 185267

llvm-svn: 185266

llvm-svn: 185263

llvm-svn: 185262

Changing the sign when comparing the base pointer would introduce all
sorts of unexpected things like:
  %gep.i = getelementptr inbounds [1 x i8]* %a, i32 0, i32 0
  %gep2.i = getelementptr inbounds [1 x i8]* %b, i32 0, i32 0
  %cmp.i = icmp ult i8* %gep.i, %gep2.i
  %cmp.i1 = icmp ult [1 x i8]* %a, %b
  %cmp = icmp ne i1 %cmp.i, %cmp.i1
  ret i1 %cmp

into:
  %cmp.i = icmp slt [1 x i8]* %a, %b
  %cmp.i1 = icmp ult [1 x i8]* %a, %b
  %cmp = xor i1 %cmp.i, %cmp.i1
  ret i1 %cmp

By preserving the original sign, we now get:
  ret i1 false

This fixes PR16483.

llvm-svn: 185259

llvm-svn: 185258

Real world code sometimes has the denominator of a 'udiv' be a
'select'.  LLVM can handle such cases but only when the 'select'
operands are symmetric in structure (both select operands are a constant
power of two or a left shift, etc.).  This falls apart if we are dealt a
'udiv' where the code is not symetric or if the select operands lead us
to more select instructions.

Instead, we should treat the LHS and each select operand as a distinct
divide operation and try to optimize them independently.  If we can
to simplify each operation, then we can replace the 'udiv' with, say, a
'lshr' that has a new select with a bunch of new operands for the
select.

llvm-svn: 185257

llvm-svn: 185251

llvm-svn: 185250

No functionality change.
It should suffice to check the type of a debug info metadata, instead of
calling Verify. For cases where we know the type of a DI metadata, use
assert.

llvm-svn: 185249

llvm-svn: 185243

We may, after other optimizations, find ourselves with IR that looks
like:

  %shl = shl i32 1, %y
  %cmp = icmp ult i32 %shl, 32

Instead, we should just compare the shift count:

  %cmp = icmp ult i32 %y, 5

llvm-svn: 185242

llvm-svn: 185241

Thename says it's an i32*, but it was actually creating another i8*

llvm-svn: 185239

llvm-svn: 185238

llvm-svn: 185237

Allow block frequencies to exceed 32 bits by using the new
BlockFrequency division function.

llvm-svn: 185236

This fixes PR16418, which reports that a function calling
__builtin_unwind_init() asserts. The cause is that this generates a
spill/restore for VRSAVE, and we support that only on Darwin (because VRSAVE is
only really used on Darwin).

The test case checks only that we don't crash. We can add correctness checks
once someone verifies what behavior the function is supposed to have.

llvm-svn: 185235

- should resolve windows buildbot failure

llvm-svn: 185232

To support this we have to insert 'extractelement' instructions to pick the right lane.
We had this functionality before but I removed it when we moved to the multi-block design because it was too complicated.

llvm-svn: 185230

Change assert("text") to assert(0 && "text").  The first case is a const char *
to bool conversion, which always evaluates to true, never triggering the
assert.  The second case will always trigger the assert.

llvm-svn: 185227

llvm-svn: 185226

- mistakenly used get_current_dir() linux function
- replaced with getcwd/_getcwd as appropriate for current platform

llvm-svn: 185225

Revising the MCJIT ObjectCache interface to allow subclasses to avoid retaining references to returned objects

llvm-svn: 185221

llvm-svn: 185219

Hopefully, this fixes the PPC64 buildbots.

llvm-svn: 185218

In this code we keep track of pointers that we are allowed to read from, if they are accessed by non-predicated blocks.
We use this list to allow vectorization of conditional loads in predicated blocks because we know that these addresses don't segfault.

llvm-svn: 185214

- lit tests verify that each line of input LLVM IR gets a !dbg node and a
  corresponding entry of metadata that contains the line number 
- unit tests verify that DebugIR works as advertised in the interface
- refactored some useful IR generation functionality from the MCJIT unit tests
  so it can be reused

llvm-svn: 185212

Patch By: Alex Deucher

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
llvm-svn: 185209