in X86-64 CCC, i8/i16 arguments are already properly zext/sext'd on input.
Capture this so that downstream zext/sext's are optimized out.  This
compiles:
  int test(short X) { return (int)X; }

to:

_test:
        movl %edi, %eax
        ret

instead of:

_test:
        movswl %di, %eax
        ret

GCC produces this bizarre code:

_test:
        movw    %di, -12(%rsp)
        movswl  -12(%rsp),%eax
        ret

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34623 91177308-0d34-0410-b5e6-96231b3b80d8

new testcase

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34622 91177308-0d34-0410-b5e6-96231b3b80d8

Fold (sext (truncate x)) more aggressively, by avoiding creation of a
sextinreg if not needed.   This is useful in two cases: before legalize,
it avoids creating a sextinreg that will be trivially removed.  After legalize
if the target doesn't support sextinreg, the trunc/sext would not have been
removed before.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34621 91177308-0d34-0410-b5e6-96231b3b80d8

track signedness of formal argument, though we have a fixme here.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34620 91177308-0d34-0410-b5e6-96231b3b80d8

Add an isNegative method to determine if the APInt's value is negative.
This is much less expensive than a test against zero.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34619 91177308-0d34-0410-b5e6-96231b3b80d8

Rewrite lshr to not do bit by bit copy but to copy and shift whole words.
This makes it much more efficient.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34618 91177308-0d34-0410-b5e6-96231b3b80d8

Fix sext operation. Shifting by zero would leave an incorrect mask.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34617 91177308-0d34-0410-b5e6-96231b3b80d8

1. Fix the flip() method to correctly flip all words of the APInt.
2. Implement the trunc, sext, and zext operations.
3. Improve fromString to accept negative values as input.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34616 91177308-0d34-0410-b5e6-96231b3b80d8

Fix an X86-64 abi bug.  We now compile:

void foo(short);
void bar(unsigned short A) {
  foo(A);
}

into:

_bar:
        subq $8, %rsp
        movswl %di, %edi
        call _foo
        addq $8, %rsp
        ret

instead of:

_bar:
        subq $8, %rsp
        call _foo
        addq $8, %rsp
        ret

Testcase here: test/CodeGen/X86/x86-64-shortint.ll

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34615 91177308-0d34-0410-b5e6-96231b3b80d8

new testcase

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34614 91177308-0d34-0410-b5e6-96231b3b80d8

fix CodeGen/X86/2007-02-25-FastCCStack.ll, a regression from my patch last
night:  fastcc returns should only go in XMM0 if we have SSE2 or above.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34613 91177308-0d34-0410-b5e6-96231b3b80d8

new testcase

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34612 91177308-0d34-0410-b5e6-96231b3b80d8

new testcase

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34611 91177308-0d34-0410-b5e6-96231b3b80d8

Fix for PR1224.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34610 91177308-0d34-0410-b5e6-96231b3b80d8

Test for PR1224.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34609 91177308-0d34-0410-b5e6-96231b3b80d8

Rework GlobalValue::removeDeadConstantUsers to always remove dead constant
exprs hanging off a global, even if the global is not otherwise dead.  This
requires some tricky iterator gymnastics.

This implements Transforms/GlobalOpt/constantexpr-dangle.ll by deleting a
constantexpr that made it appear that the address of the function was taken.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34608 91177308-0d34-0410-b5e6-96231b3b80d8

new testcase.  @foo should be marked fastcc by globalopt

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34607 91177308-0d34-0410-b5e6-96231b3b80d8

disable some noisy debug output

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34606 91177308-0d34-0410-b5e6-96231b3b80d8

no really, this is the right patch

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34605 91177308-0d34-0410-b5e6-96231b3b80d8

always promote float varargs to double.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34604 91177308-0d34-0410-b5e6-96231b3b80d8

1. Provide more detail in file comment.
2. Move comments for methods to .h file, delete them in .cpp file.
3. All places that were doing manual clear of high order bits now call the
   clearUnusedBits() method in order to not depend on undefined behavior
   of the >> operator when the number of bits shifted equals the word size.
4. Reduced # of loc by using the new result of clearUnusedBits() method.
5. Simplified logic (decreased indentation) in a few places.
6. Added code comments to larger functions that needed them.
7. Added FIXME notes about weak implementations of things (e.g. bit-by-bit
   shift right is sub-optimal).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34603 91177308-0d34-0410-b5e6-96231b3b80d8

Fix clearUnusedBits to not depend on "undefined behavior" of >> operator
when the bit size is equal to the word size. This happens to work out okay
on x86, but might not on other platforms. The change just detects when
there are no bits to clear (because BitWidth is a multiple of the word size)
and returns early.

Also, move some comments from .cpp file into header.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34602 91177308-0d34-0410-b5e6-96231b3b80d8

one important bugfix: PPC32 didn't have both elf and macho support for
external symbols and global addresses.  Add the missing ones.

one important workaround: PPCISD::CALL is matched by both PPCcall_ELF
and PPCcall_Macho, disable the _ELF patterns for now.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34601 91177308-0d34-0410-b5e6-96231b3b80d8

add -enable-eh

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34600 91177308-0d34-0410-b5e6-96231b3b80d8

optimize duplicate ValueMap lookups

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34599 91177308-0d34-0410-b5e6-96231b3b80d8

cosmetic changes from review of last patch.  obvious

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34598 91177308-0d34-0410-b5e6-96231b3b80d8

A couple of more places where a register liveness has been extended and its last kill should be updated accordingly.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34597 91177308-0d34-0410-b5e6-96231b3b80d8

Add an assertion.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34596 91177308-0d34-0410-b5e6-96231b3b80d8

Fix a couple of bugs related IsDead back propagation during coalescing.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34595 91177308-0d34-0410-b5e6-96231b3b80d8

If the liveinterval of the source instruction has been extended, remove the IsKill marker.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34594 91177308-0d34-0410-b5e6-96231b3b80d8

Only add liveinterval to livein set if it isn't assigned a stack slot.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34593 91177308-0d34-0410-b5e6-96231b3b80d8

fastcc functions that return double values now return them in xmm0 on x86-32.
This implements CodeGen/X86/fp-stack-ret.ll:test[23]

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34592 91177308-0d34-0410-b5e6-96231b3b80d8

verify that double is returned in XMM0 if the function is fastcc.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34591 91177308-0d34-0410-b5e6-96231b3b80d8

allow vectors to be passed to stdcall/fastcall functions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34590 91177308-0d34-0410-b5e6-96231b3b80d8

move LowerRET into the 'Return Value Calling Convention Implementation'
section of the file.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34589 91177308-0d34-0410-b5e6-96231b3b80d8

make all Lower*CallTo implementations use LowerCallResult to handle their
result value stuff.  This eliminates a bunch of duplicated code and now
GetRetValueLocs is the sole place that decides where a value is returned.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34588 91177308-0d34-0410-b5e6-96231b3b80d8

pass the calling convention into Lower*CallTo, instead of using ad-hoc flags.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34587 91177308-0d34-0410-b5e6-96231b3b80d8

factor a bunch of code out of LowerCCCCallTo into a new LowerCallResult
function.  This function now uses GetRetValueLocs to determine *where*
the result values are located and concerns itself with *how* to pull the
values out.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34586 91177308-0d34-0410-b5e6-96231b3b80d8

move some code around, pass in calling conv, even though it is unused

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34585 91177308-0d34-0410-b5e6-96231b3b80d8

fold trivial token factor nodes.  This allows us to compile
test/CodeGen/X86/fp-stack-ret.ll into:

        movl 4(%esp), %eax
        fldl (%eax)
        ret

instead of:

        subl $12, %esp
        movl 16(%esp), %eax
        movsd (%eax), %xmm0
        movsd %xmm0, (%esp)
        fldl (%esp)
        addl $12, %esp
        ret

by eliminating a token factor that blocked a check.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34584 91177308-0d34-0410-b5e6-96231b3b80d8

new testcase

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34583 91177308-0d34-0410-b5e6-96231b3b80d8

simplify result value lowering by splitting the selection of *where* to return
registers out from the logic of *how* to return them.

This changes X86-64 to mark EAX live out when returning a 32-bit value,
where before it marked RAX liveout.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34582 91177308-0d34-0410-b5e6-96231b3b80d8

Allow this to compile now that the header file is checked in.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34581 91177308-0d34-0410-b5e6-96231b3b80d8

Add a private constructor for efficiency.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34580 91177308-0d34-0410-b5e6-96231b3b80d8

make void-return not a special case

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34579 91177308-0d34-0410-b5e6-96231b3b80d8

eliminate a bunch more temporary vectors from X86 lowering.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34578 91177308-0d34-0410-b5e6-96231b3b80d8

eliminate temporary vectors created during X86 lowering.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34577 91177308-0d34-0410-b5e6-96231b3b80d8

remove std::vector's in RET lowering.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34576 91177308-0d34-0410-b5e6-96231b3b80d8

verify i128 return on x86-64 continues to codegen optimally.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34575 91177308-0d34-0410-b5e6-96231b3b80d8

implement support for the linux/ppc function call ABI.  Patch by
Nicolas Geoffray!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34574 91177308-0d34-0410-b5e6-96231b3b80d8

Add XLForm_1_ext template, patch by Nicolas Geoffray.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34573 91177308-0d34-0410-b5e6-96231b3b80d8

Improve JIT support for linux/ppc: Patch by Nicolas Geoffray!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34572 91177308-0d34-0410-b5e6-96231b3b80d8

this doesn't compile, disable it

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34571 91177308-0d34-0410-b5e6-96231b3b80d8

remove these.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34570 91177308-0d34-0410-b5e6-96231b3b80d8

Clean up lshr and ashr to coding standards.
Handle the single word cases for shiftAmt == BitWidth.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34569 91177308-0d34-0410-b5e6-96231b3b80d8

remove crediting of Evan Cheng and me.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34568 91177308-0d34-0410-b5e6-96231b3b80d8

initialize a instance variable

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34567 91177308-0d34-0410-b5e6-96231b3b80d8

update this for llvm-gcc4

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34566 91177308-0d34-0410-b5e6-96231b3b80d8

Whoops, last word with bits in large shift left wasn't correct.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34565 91177308-0d34-0410-b5e6-96231b3b80d8

Fix the > 64 bits case for left shift.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34564 91177308-0d34-0410-b5e6-96231b3b80d8

Removed WaterListOffset, inserted BBOffsets.  Remove TODO item about this
from README.
When no water available, use end of block if in range.  (More to do here.)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34563 91177308-0d34-0410-b5e6-96231b3b80d8

Fix the remainder shifting in KnuthDiv.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34562 91177308-0d34-0410-b5e6-96231b3b80d8

1. Fix a bug in fromString for the <= 64bits case
2. Fix shl when shiftAmount == BitWidth.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34560 91177308-0d34-0410-b5e6-96231b3b80d8

1. Fix last bug in KnuthDiv. All divide tests pass up to 1024 bits now.
2. Clean up comments, style, coding standards, etc.
3. Simplify a constructor.

Extended testing revealed some additional bugs in shifting. I'll fix these
tomorrow.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34559 91177308-0d34-0410-b5e6-96231b3b80d8

Improve documentation.
Make divide function internal (it was briefly external for testing).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34557 91177308-0d34-0410-b5e6-96231b3b80d8

Deal with cases when MMI is not requested.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34556 91177308-0d34-0410-b5e6-96231b3b80d8

Drop unused operand.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34555 91177308-0d34-0410-b5e6-96231b3b80d8

1. Make internal functions take const arguments where they should, just
   to be safe.
2. Make internal functions that return a carry/borrow return bool instead
   of uint64_t because the carry/borrow can only be in range [0,1].
3. Assert that the pointers to KnuthDiv are all different so that the
   result and operands can't overlap.
4. Add debug output to KnuthDiv function.
5. Fix a problem with KnuthDiv by separating the b's complement operation
   from the subtraction borrow operation. This fixes a wide range of
   division problems, but alas, not all of them.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34554 91177308-0d34-0410-b5e6-96231b3b80d8

Added some more information on how to use "delta" to reduce testcases.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34553 91177308-0d34-0410-b5e6-96231b3b80d8

Fix PR1216 by cleaning up the ownership of JITResolver.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34552 91177308-0d34-0410-b5e6-96231b3b80d8

Fix CodeGen/Generic/2007-02-23-DAGCombine-Miscompile.ll and PR1219

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34551 91177308-0d34-0410-b5e6-96231b3b80d8

testcase for PR1219

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34550 91177308-0d34-0410-b5e6-96231b3b80d8

remove folding set debug output

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34549 91177308-0d34-0410-b5e6-96231b3b80d8

Fix Transforms/ConstProp/2007-02-23-sdiv.ll and PR1215

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34548 91177308-0d34-0410-b5e6-96231b3b80d8

testcase for pr1215

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34547 91177308-0d34-0410-b5e6-96231b3b80d8

Fix InstCombine/2007-02-23-PhiFoldInfLoop.ll and PR1217

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34546 91177308-0d34-0410-b5e6-96231b3b80d8

testcase for pr1217

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34545 91177308-0d34-0410-b5e6-96231b3b80d8

Refactor the setName stuff, moving it down the inheritance hierarchy, to
solve a crash in -instcombine -debug that was hit while investigating PR1217

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34544 91177308-0d34-0410-b5e6-96231b3b80d8

Make the testcase correct.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34543 91177308-0d34-0410-b5e6-96231b3b80d8

Testcase for PR1164

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34542 91177308-0d34-0410-b5e6-96231b3b80d8

PR1164:
Generate local names with a "llvm_cbe_" prefix using the actual name of the
variable instead of a temporary name.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34540 91177308-0d34-0410-b5e6-96231b3b80d8

Avoid TBAA issue.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34539 91177308-0d34-0410-b5e6-96231b3b80d8

New test - practically wrote itself.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34538 91177308-0d34-0410-b5e6-96231b3b80d8

bugfix: SP wasn't updated for varargs when frame pointer was eliminated.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34537 91177308-0d34-0410-b5e6-96231b3b80d8

Reuse extends the liveness of a register. Transfer the kill to the operand that reuse it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34536 91177308-0d34-0410-b5e6-96231b3b80d8

Handle improper cast.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34535 91177308-0d34-0410-b5e6-96231b3b80d8

Handle cases when joining live intervals of two virtual registers.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34534 91177308-0d34-0410-b5e6-96231b3b80d8

Don't spill LR as a callee saved register.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34533 91177308-0d34-0410-b5e6-96231b3b80d8

Fix stack alignment in functions with varargs.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34532 91177308-0d34-0410-b5e6-96231b3b80d8

fix an obscure and tricky bug the inliner can hit sometimes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34531 91177308-0d34-0410-b5e6-96231b3b80d8

fix this testcase

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34530 91177308-0d34-0410-b5e6-96231b3b80d8

crtend is only for llvm-gcc3, remove it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34529 91177308-0d34-0410-b5e6-96231b3b80d8

Fix thinko.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34528 91177308-0d34-0410-b5e6-96231b3b80d8

Loop passes are set up to accept pointer.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34527 91177308-0d34-0410-b5e6-96231b3b80d8

Mark livein registers as being in used.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34525 91177308-0d34-0410-b5e6-96231b3b80d8

clarify.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34524 91177308-0d34-0410-b5e6-96231b3b80d8

rewrite of constant islands

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34523 91177308-0d34-0410-b5e6-96231b3b80d8

-march=arm -enable-thumb => -march=thumb

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34522 91177308-0d34-0410-b5e6-96231b3b80d8

Added -march=thumb; removed -enable-thumb.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34521 91177308-0d34-0410-b5e6-96231b3b80d8

80 col. violation.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@34520 91177308-0d34-0410-b5e6-96231b3b80d8